energy-aware ant colony optimization based routing for mobile ad ...
Reliable Ant Colony Routing Algorithm for Dual-Channel ...
Transcript of Reliable Ant Colony Routing Algorithm for Dual-Channel ...
Research ArticleReliable Ant Colony Routing Algorithm forDual-Channel Mobile Ad Hoc Networks
YongQiang Li12 Zhong Wang 1 QingWen Wang 1
QingGang Fan 1 and BaiSong Chen1
1Xirsquoan Research Institute of High Technology Xirsquoan 710025 China2Science and Technology on Communication Networks Laboratory Shijiazhuang 050000 China
Correspondence should be addressed to QingWenWang wqw013890163com
Received 2 March 2018 Accepted 11 April 2018 Published 15 May 2018
Academic Editor Ximeng Liu
Copyright copy 2018 YongQiang Li et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
For the problem of poor link reliability caused by high-speed dynamic changes and congestion owing to low network bandwidthin ad hoc networks an ant colony routing algorithm based on reliable path under dual-channel condition (DSAR) is proposedFirst dual-channel communication mode is used to improve network bandwidth and a hierarchical network model is proposed tooptimize the dual-layer networkThus we reduce network congestion and communication delay Second a comprehensive reliablepath selection strategy is designed and the reliable path is selected ahead of time to reduce the probability of routing restart Finallythe ant colony algorithm is used to improve the adaptability of the routing algorithm to changes of network topology Simulationresults show that DSAR improves the reliability of routing packet delivery and throughput
1 Introduction
Ad hoc networks do not rely on communication infrastruc-tures Their nodes have the simultaneous function of routingand terminal computing This creates a multihop dynamicwireless network quickly via self-organization [1] In recentyears ad hoc networks have been widely used in disasterrescue vehicular networks battlefields remote mountainareas fire zones earthquake scenarios multihop satellite net-works and underwater acoustic sensor networks [2] Owingto the dynamic changes of network topology caused by thehigh-speed movement of nodes the exhaustion of batteryenergy and the multihop characteristics of ad hoc networksrouting links are broken easily Thus the reliability of thenetwork is reduced The key to improving the reliability inmobile ad hoc networks is selecting reliable routing andnodes with large remaining energy Some routing protocolssuch asAODV[3] andLOADng [4] use the hopsmechanismHowever when nodes move quickly the path may be brokeneasily To ensure reliable data packet transmission and reduceenergy consumption a novel reliability prediction model isproposed The ant colony algorithm is applied to select areliable path DSAR takes the integrated pathrsquos reliable value
as the routing standard which can better adapt to the changesof network topology
In mobile wireless ad hoc networks when the commu-nication load and the number of nodes increase the single-channel technology will cause network capacity limitationsand performance degradation The multichannel mode hasthe advantages of low latency and high capacity [5]Thus thedual-channel technology is adopted in this paper
The major contributions of this paper are as follows
(i) We propose a novel dual-channel communicationmode to separate the control and data layers
(ii) We provide a novel metric to measure the reliabilityof the integrated link and node
(iii) A heuristic technique is adopted in DSAR to find theoptimal and most reliable route faster than existingschemes
(iv) Both forward ant and backward ant simultaneouslyupdate via pheromones
(v) The significance of DSAR is evaluated by comparingsimulation results with AODV and EEABR
HindawiWireless Communications and Mobile ComputingVolume 2018 Article ID 4746020 10 pageshttpsdoiorg10115520184746020
2 Wireless Communications and Mobile Computing
The paper is designed as follows The related work of thepaper is in Section 2 Section 3 proposes an ant colony routingalgorithm with reliability prediction under dual-channelconditions Section 4 proposes a route discovery processbased on joint optimization of dual-channel networks andSection 5 reveals simulation results and discussions FinallySection 6 concludes this paper
2 Related Work
Routing protocols designed for wireless ad hoc networksface many challenges especially in highly dynamic networksMany researchers have proposed algorithms to address this[6ndash13] These routing protocols can be divided into threecategories based on the different route discovery strategiesproactive on-demand and hybrid protocols In the activerouting protocols (eg DSDV [11] FSR [12] GSR [14] andHSR [6]) each network node periodically exchanges routinginformation with other nodes and each node must save therouting table whereas the proactive routing protocol hasa low delay it saves unnecessary routing information andconsumesmanynetwork resourcesWith on-demand routingprotocols (eg DSR [15] AODV [3] RDMAR [16] and LAR[17]) the routing information is created only when nodesare needed However there is a large delay with this type ofprotocol Both proactive and on-demand routing protocolscannot completely solve routing problems Therefore manyscholars have proposed hybrid routing protocols that com-bine the features of active and on-demand routing (eg ZRP[18] HWMP [19] and AOHR [20])
21 Representative Schemes for Comparison In the past fewyears swarm intelligence algorithms have become very popu-lar with ad hoc networks especially the ant colony optimiza-tion (ACO) algorithm The ACO algorithm is based on theantsrsquo foraging behavior in nature to find the optimal pathGraphically it has been applied widely in various fields [21]In the ant colony algorithm the intelligence of a single ant islimited but the ants can accomplish complex tasks via groupcooperation ACO has the same characteristics as the routingdesign of mobile ad hoc networks [22] G Dicaro proposedACO via AntNet [23] AntNet is based on the wired networkand the hop mechanism for data transmission When it istransplanted into ad hoc networks the power consumption ofsome nodes grows too high leading to unstable links Thusit is not suitable for mobile ad hoc networks
ARA protocol [24] proposed by Gunes et al in 2002 wasthe earliest application of ACO in ad hoc networks Howeverthe ARA protocol does not consider the energy balance ofnodes
Correia and Vazao proposed the SARA protocol [25]which was improved based on the ARA protocol SARA usesa restricted neighbor broadcast mechanism in which eachnode broadcasts the ant to all the neighbor nodes but onlyone neighbor is selected to send the forward ant SARA usesthe same routing maintenance and routing error handlingmechanism as the ARA protocol With routing maintenanceto reduce routing consumption the data packets also updatepheromones on the link During routing repair the depth
search algorithm is used to control the number of nodes torepair the routing and reduce the routing overhead caused bythe source node restart routing discovery process Comparedto theARAalgorithm the SARAalgorithmhas high through-put and low routing overhead but the route establishmentprocess of SARA takes a long time
AntHocNet [26] is a hybrid algorithm that combines theadvantages of AntNet and ARA protocols The routing pro-cess is carried out on demand and the routing mainte-nance process is carried out actively Compared with AODVthe packet delivery rate of AntHocNet protocol has beenimproved significantly However because of the active rout-ing maintenance mechanism the control overhead is muchhigher than AODV
The ACECR algorithm [27] was proposed by Zhou et alACECR is based on the ACO algorithm whose pheromoneupdates depend on two aspects the number of hops and theremaining energy However this algorithm does not considerstability and reliability
The R-ACO algorithm [28] preselects highly stable linksfor packet sending as far as possible to avoid the low linkstability of node processing packets Compared to LAR andAntHocNet R-ACO improves the success rate of data trans-mission while reducing communication overhead HoweverR-ACO algorithm increases the length of the path andrequires nodes to carry GPS devices
In [2] the level of pheromone is determined accordingto the routing length its congestion and the end-to-end pathreliabilityThis protocol provides high data delivery rates withlow end-to-end delay
For AOCR [29] the authors proposed an on-demand antcolony clustering routing protocol based on a weakly con-nected dominating set (AOCR) AOCR adopts the weaklyconnected dominating set as the auxiliary structure of cluster-ing nodesThe forward ants are only broadcast by the heads ofeach clusterThepheromone intensity depends on the averageresidual energy of the network and the minimum energyvalue of the nodes on the path AOCR adopts a pseudoran-dom proportional rule to select the most efficient routingCompared to other ant colony protocols AOCR requires lessstorage space and fewer network transmission resources toperform intelligent routing search
The authors in [30] proposed FTAR which avoids for-warding nodes with erroneous tendencies thus improvingthe performance of the network Compared with the DSRalgorithm E2FT [31] and AntHocNet FTAR improves datatransmission success rate and throughput However FTARuses an iterative method to obtain path confidence whichincreases the processing time of packets
The authors in [32] proposed an ant colony optimizationrouting algorithm (ie POSANT) based on geographic loca-tion information For the problem of the excessive numberof control packets and high transmission delays in ACOPOSANT combined the advantages of ACO and geographiclocation information Therefore it reduced the time of dis-covering routing and the number of ants generated by usinglocation information Compared to the GPSR algorithmPOSANT reduced the delay and increases the success rate of
Wireless Communications and Mobile Computing 3
data transmission but it did not consider the reliability of thepath
The authors in [33] proposed an enhanced congestioncontrol multipath routing method with ACO optimizationfor ad hoc networks which addressed the problem of linkblockage Additionally the load rapidly increases on the linkThey proposed an ACO-based multipath congestion controltechnique that varies the queue according to the load ina dynamic network Simulation showed that the proposedACO protocol had good performance However the algo-rithm did not use multiple channels and does not fundamen-tally solve the problem of network congestion
The authors in [5] presented a dual-channel network clus-tered routing protocol (DNCRP) a hybrid routing protocolDNCRP uses both 2-hop-distance neighbor cluster IDs andnode attribution information to search the path of interclusterrouting within the 2-hop-distance neighbor clusters perthe source cluster Simulation results show that DNCRP issuitable for high mobility networks
In [34] the energy of nodes was considered as a prior fac-tor for route choice However this scheme does not considerlink reliability as a factor for route choice and increase end-to-end delay
In summary there are two major limitations of the cur-rent routing algorithm First it does not use dual-channelstrategies Second there are very few routing schemes thatconsider the reliability of the integrated nodes and links as aprior factor for route-choosing
Hence there is a need to develop a unified routing proto-col which can fulfill the low end-to-end delay low routingoverhead and high reliability In this paper a reliable antcolony algorithm based on dual-channel conditions (DSAR)is proposed for ad hoc networks DSAR overcomes alllimitations of the previous schemes
3 An Ant Colony RoutingAlgorithm with Reliability Prediction underDual-Channel Conditions
31 Dual-Channel Joint Optimization Model In ad hoc net-works owing to the limited bandwidth of the nodes the datatransmission delay is high With technological developmenta node in ad hoc networks can be configured with twochannels which can reduce collisions increase bandwidthease network congestion and improve network performance[35] The dual-channel network communication model issimplified to the routing problem with a hierarchical mapwithout considering the channel assignment problem in thenetwork [36] In this paper we use a dual-channel layeredtransmission mode one channel as the control layer andanother as the data layer The control packets are transmittedin the control layer and the data packets are transmitted inthe data layer This double channeling eliminates messageconflict and reduces the delay of channel handoff If thecontrol layer is congested and the data layer has enoughbandwidth resources the dual-channel joint optimizationmode can transfer control packets in the control layer to thedata layer in real time to complete the joint scheduling of
the double-layer network and reduce congestion The dual-channel model is shown in Figure 1
32 Basic Ant Colony Algorithm for Ad Hoc Networks Whenan ant walks into an intersection it randomly selects a paththat has not been passed and releases pheromones Thesize of the pheromone is related to the path length Thelonger the path the smaller the pheromone When ants passby this intersection they will choose the path where thepheromone is large Thus a positive feedback is formedand the pheromone quantity on the optimal path is largerand the pheromone on other paths will become less overtime Simultaneously the whole ant colony can adapt tothe change of environment When ants suddenly encounterobstacles along the way they can quickly adjust their pathThus in the process of the entire colony finding the antsrsquopath a single antrsquos optimal path selection ability is limitedbut the ant colony has good self-organization because of theglobal pheromone Sharing path information the ants findthe optimal path via collective behavior of the ant communityThe ant colony algorithm has distributed parallel computercontrol which is easy to combine with other algorithms andhas strong robustness
The ant colony optimization algorithm has been success-fully applied to many optimization combinatorial problems[37] The ant foraging process is very similar to the routingproblem of ad hoc networks In this paper the nest and foodare compared to the source node and the destination nodein ad hoc networks The ant colony algorithm uses an antdecision table which comprises a node selection probabilityfrom a path and relevant local information The ants usethis decision table to guide their search of the mobile spacein the optimal region which is the process of forming therouting tableThus the ant colony algorithmcan be used in adhoc networks Through the pheromone mechanism the antssearch for and maintain optimal routing The mechanism ofevaporation updates the pheromone of each node which canquickly adapt to the needs of the dynamic changes of ad hocnetworks
In these networks the network topology model is thewireless graph 119866(119881 119864) where 119881 is a network node and 119864 isthe link between two nodes At time 119905 there are 119886119894(119905) antsThetotal number of ants in the network is119898 = sum119899119894=1 119886119894(119905) 119875119894119895(119905) isthe probability of choosing link 119864119894119895 for ant119870 at time 119905
119875119896119894119895 (119905) =
[120591119894119895 (119905)]120572 [120578119894119895]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895]120573
119895 isin allowed119896
0 else
(1)
where 120591119894119895(119905) is the strength of the pheromone in the link 119864119894119895120572 is a parameter to measure the trajectory of pheromones120578119894119895 is visibility between node 119894 and node 119895 which is generallydefined as 1119889119894119895 (119889119894119895 is the distance between node 119894 and node119895) 120573 is a parameter that measures visibility and allowed119896 is acollection of nodes that have not been visited
4 Wireless Communications and Mobile Computing
LL
InterfaceQueue
MAC
Networkinterface
LL
InterfaceQueue
MAC
ARP ARP
Application
RoutingAgent
Channel 1
Networkinterface
Channel 2
PropagationMode
Addr
ess
Mul
tiple
xer Po
rtM
ultip
lexe
r
Figure 1 Dual-channel joint optimization mode
The pheromone update formula on each path in ad hocnetworks is as follows
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) + Δ120591119894119895 (2)
where 120588 is the pheromone volatilization coefficient which isa constant between 0 and 1 and Δ120591119894119895 is the increment of thepheromone of ants passing through links 119894 and 119895
Δ120591119894119895 = 119897sum119896=1
Δ120591119896119894119895 (3)
33 Route Reliability In ad hoc networks there are twomainreasons for path breaking One is the movement of nodeson the communication path and the other is the nodeswithdrawing from the network because of energy depletionThus we select relatively reliable nodes and links Then thepath stability (PS) factor is introduced to judge the stability ofthe path During the establishment of QoS routing the pathwith the strongest stability is selected from themultiple pathssatisfying the QoS requirements reducing the probabilityof path breaking In this paper the path stability factor isthe function of the link stability factor and the node energystability factor
331TheNode Energy Stability Factor Suppose that there are119895 intermediate nodes in the path 119875119894 119873119894 = 1198991198941 1198991198942 119899119894119895
In this paper we define the node energy stability fac-tor
ES119894119895 = 119864currrent (119894119895)119864initial (119894119895) (4)
where 119864initial(119894119895) is the initial energy of node 119895 in path 119875119894 and119864currrent(119894119895) is the current remaining energy of node 119895 in path119875119894Because a node in the path cannot be used owing to the
exhausted energy the energy stability factor of the path is theminimum energy stability factor of all nodes in the path 119875119894
ES119894 = min ES119894119895 119897119894119895 isin 119901119894 (5)
332 The Link Stability Factor In this paper we define thelink reliability factor as the remaining lifetime of the linkThe communication radius of each node is 119877 Each nodeis equipped with GPS thus every node can perceive thelocation speed time of nodes and period and send itsown coordinates and speed information to neighbor nodesAccording to the location information of nodes the remain-ing lifetime of each link can be predicted and the link stabilityfactor can be obtained As shown in Figure 2 the initialdistance between the two nodes119872 and119873 is 119889
The relative motion of the two nodes is equivalent toone node moving while the other node is stationary Thecoordinates of node119873 relative to the stationary node119872 are
Wireless Communications and Mobile Computing 5
M
N
R
MN
Figure 2 Calculate the lifetime of the link119872119873
(119909119873 minus 119909119872 119910119873 minus 119910119872) Thus the distance between119872 and119873 is119889 = radic1199091198721198732 + 1199101198721198732 According to the relative movement ofthe two nodes after time 119905 node119872 relative to position119873 is(1199091015840119872119873 1199101015840119872119873)
1199091015840119872119873 = 119909119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 cos 120579119872119873
1199101015840119872119873 = 119910119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 sin 120579119872119873 (6)
When the distance between 119872 and 119873 is 119877 the link between119872 and119873 is broken
1198891015840 = radic(1199091015840119872119873)2 + (1199101015840119872119873)2 = 119877 (7)
Take (6) into (7)
10038161003816100381610038161003816997888997888997888V119873119872100381610038161003816100381610038162 1199052 + 2 10038161003816100381610038161003816997888997888997888V119873119872
10038161003816100381610038161003816 (119909119873119872 cos 120579119873119872 + 119910119873119872 sin 120579119873119872) 119905+ 1198892 minus 1198772 = 0 (8)
where997888V119872 = (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894 + (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 997888119895 997888V119873 = (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873) 997888119894 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873) 997888119895 997888997888997888V119873119872 = 997888V119873 minus 997888V119872= (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894
+ (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 99788811989510038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic(10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)2 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872)2120579119873119872 = tanminus1(
10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 12057911987210038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)
(9)
Because 119905 cannot be negative 119905 becomes the following
119905 = radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)210038161003816100381610038161003816997888997888997888rarrV11987311987210038161003816100381610038161003816
minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
(10)
Link stability is
LS119872119873 = 119905119905max= 119905
119877 10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)2119877minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)119877
(11)
There is a 119896 link in a path between the source node 119878 and thedestination node 119863 The link stability of the path dependson the minimum link stability factor in the 119896-segment linkThus we define the link reliability factor of path 119875119894 as theminimum link stability factor
LS119894 = min LS119872119873 119897119872119873 isin 119901119894 (12)
333 The Path Reliability Factor Considering the link stabil-ity factor and node energy stability factor the comprehensivereliability factor of the path is defined as
PS119894 = 11ES119894 + 1LS119894 =ES119894 times LS119894ES119894 + LS119894
(13)
Whether the communication path between the source nodeand the destination node is stable depends on the worst linksand nodes in the path because the entire communication isinterrupted when there is a broken link or when a node exitsthe network because of exhausted battery energy After usingthe abovemethod if either of the two stability factors is smallthe value of PS119894 will be small Thus path reliability will bepoor When the source node finds multiple paths satisfyingthe QoS requirements the one with the largest PS is selectedThus PS = max PS119894 119875119894 isin 1198754 Route Discovery Process Based on JointOptimization of Dual-Channel Networks
To improve channel utilization based on the double-channelmodel and ant colony optimization algorithm the two-interlayer joint optimization routing mode is proposed Itcan increase bandwidth and make full use of idle resourcesbetween different layers In Mode 1 the routing service in thecontrol layer can only be transmitted in the control layer Ifthe control layer does not have enough channel resourcesthe service will be rejected To reduce the blocking rate of thecontrol layer joint optimization Mode 2 is proposed Whenthe data layer has enough idle resources the control packetsin the control layer can be transmitted to the data layer inreal time to realize the joint optimization of the two-layernetwork The specific routing process of the two modes is asfollows
Mode 1 It is the route discovery mode on the control layer ofDSAR
(1) Broadcast a Hello Message At the control layer eachnetwork node periodically sends a 1-hop hello message The
6 Wireless Communications and Mobile Computing
Table 1 The format of the hello message
Type Src addr Scr 119883 Scr 119884 Scr 119881 119873 energy
Table 2 The structure of neighbor
119873 addr 119873 energy LS Hops119873 119883 Nr 119884 119873 119881 Phenomenon
format of the hello message is shown in Table 1 whereldquoTyperdquo denotes packet type ldquoSrc addrrdquo is the address of thesource node that sends the hello message ldquoScr Xrdquo are the Xcoordinates of node ldquoScr Yrdquo are the Y coordinates of nodeldquoScr Vrdquo is velocity vector of node and ldquoN energyrdquo is residualenergy of node
(2) Establish the Neighbor List Each node establishes itsneighbor list by receiving hello messages sent by neighbornodes in real time In this paper each network node hasa GPS positioning device to obtain its geographic locationinformation Each network node sends hello packets period-ically and accepts those sent by neighbor nodes Each hellopacket contains node coordinates and speed information andthe neighbor list of each node The neighbor list contains thelocation vector information of all neighboring nodes Thuseach network node can obtain the location information ofneighbor nodesThe frequency of packet transmission can beset according to differentmotion scenarios and the transmis-sion frequency of hello messages can be higher for scenarioswhere network topology changes rapidly In this paper wedesign hellomessaging with a sending interval of 1 s If a nodedoes not receive neighbor hello packets in 1 s the link of thisnode is broken and the node deletes it from its neighbortable The node updates its neighbor list immediately afterthe hello message is received Otherwise the comprehensivereliability of the node is calculated according to formula (13)The format of neighbor message is shown in Table 2 whereldquo119873 addrrdquo is the address of the source node that sends thehello message ldquo119873 energyrdquo is residual energy of neighbornode ldquo119873 119883rdquo are the119883 coordinates of neighbor node ldquo119873 119884rdquoare the 119884 coordinates of neighbor node 119873 119881rdquo is velocityvector of neighbor node ldquoLSrdquo is the stability of neighbornode ldquoHopsrdquo denotes total hops of a route passing throughthis neighbor node and ldquoPhenomenonrdquo denotes the value ofpheromone
(3) Send a Forward Ant and Update Pheromone When thesource node 119878 has data packets sent to the destinationnode119863 the source node 119878 looks at the routing informationtable namely the pheromone table If there is no routinginformation and the control layer has enough bandwidththe forward ant is broadcast in the control layer If there isa route to node119863 the packet is sent directly to the data layerIn the initial stage of routing establishment the source nodebroadcasts a certain number of forward ants at the controllayer The unicast or broadcast of each intermediate nodebetween source node 119878 and destination node 119863 depends onwhether the intermediate node has pheromone If there is
Table 3 The format of forward ant message
Type Fant addr Fant Seqno TTLScr 119884 Scr 119881 119864sum Fd119896
Seqno 0Seqno 1Seqno 119899
pheromone at each intermediate node the probability thatant 119870 selects the next hop neighbor node 119895 is calculatedaccording to the following equation
119875119896119894119895 (119905)
=
[120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573
119895 isin allowed119896
0 else(14)
where 119875119896119894119895(119905) is the probability that ant 119870 selects the next hopneighbor node 119895 when node 119894moves toward the destinationnode 120591119894119895(119905) is the size of the pheromone of node 119894 at time119905 in link 119894119895 120578119894119895(119905) is a visual function of node 119894 at time 119905from node 119894 to node 119895 allowed119896 is a collection of neighbornodes of node 119894 120572 and 120573 are adjustment coefficients 120572 is arelative importance coefficient of the residual pheromone 120573is a relative importance coefficient of heuristic informationThe pheromone of node is updated as equation
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) (15)
The format of the forward ant message is shown inTable 3 where ldquoTyperdquo denotes the packet type ldquoFant addrrdquois the address of source node ldquoFant Seqnordquo is the sequencenumber of forward ants generated by source nodes ldquoSeqnordquodenotes node address visited by the forward ant ldquoTTLrdquo is thesurvival time of forward ant and ldquoFd119896rdquo denotes the numberof hops that the forward ant moves to the current node
(4) Send a Backward Ant and Update Pheromone When aforward ant arrives at the destination node it turns into abackward ant The backward ant returns to the source nodealong its former path If a fault occurs in the link of thenext hop caused by the movement of a node in the path tothe source node then the backward ant will be discardedWhen the backward ant returns to node 119894 from node 119895 thepheromone of node 119894 is updated according to the pheromoneupdate of (16)
120591119894119895 (119905 + 1) = 120591119894119895 (119905)1 minus 120588 + Δ120591119894119895 (119905) (16)
Δ120591119894119895 (119905) = 119898sum119896=1
Δ120591119896119894119895 (119905) (17)
Wireless Communications and Mobile Computing 7
Table 4 The format of backward ant message
Type Visitednode 119864min 119864avg Bd119896
Δ120591119896119894119895 (119905) = PS119894 (18)
120578119896119894119895 (119905) = 1Fd119896
(19)
The format of the backward ant message is shown inTable 4 where ldquoTyperdquo denotes the packet type ldquoVisitednoderdquodenotes the ID of node visited ldquo119864minrdquo is the energy value ofthe minimum energy node on the path that the backward antpasses through ldquo119864avgrdquo is the residual average energy of theant 119870 to the current node and ldquoBd119896rdquo denotes the number ofhops experienced by the backward ant 119870 to node 119895Mode 2 It is the route discovery process based on jointoptimization of dual-channel networks
When the control layer is congested joint optimizationMode 2 is proposed That is when the data layer has enoughidle resources the control packets in the control layer can betransmitted to the data layer in real time to realize the jointoptimization of the two-layer network The specific routingprocess of Mode 2 is as follows
(1) If the control layer has enough network resourcesthe forward ant is routed via Mode 1 in the control layerOtherwise turn to (2)
(2) When the forward ant119870 moves to node 119894 in the datatransport layer the forward ant 119870 looks at whether thereare available channel resources for ants to find paths withneighboring nodes of 119894 If not ant 119870 stops and refuses toperform the routing lookup service Otherwise it turns to (3)
(3) Ant 119870 performs the routing service in the data trans-mission layer and finds the next hop node 119895 in Mode 1
(4)When reachingnode 119895 ant119870first investigateswhetherthere are enough channel resources between node 119895 and itsneighbor nodes in the control layer to perform the routingservice If not ant 119870 continues to perform the path-findingservice in the data transport layer Otherwise ant 119870 returnsto the control layer and searches the optimal path of serviceaccording to Mode 1
5 Simulation and Analysis
In this paper to verify the reliability of the DSAR protocolNS-2 is selected and the DSAR algorithm is compared tothe EEABR algorithm [34] and the AODV algorithm AODVis a classic routing algorithm and EEABR is a successfulapplication of the ant colony algorithm in wireless ad hocnetworks
51 Simulation Setup In a wireless simulation environmenteach mobile network node is randomly distributed in the1000m times 1000m area 50 nodes are randomly arrangedaccording to the random way-point model The communi-cation radius of each node is 250m MAC layer adopts dual-channel mode The packet length is 512 b and the send rate
varies from 1 to 16 packetss The evaporation of pheromoneoccurs every 1 s The evaporation rate 120588 is set to 02 Eachvalue of 120572 and 120573 is set to 20 and 15 respectively Thesimulation time is 1000 s To reduce random errors theexperimental results will be the average of the 10 experimentsSimulation algorithm routing layers are (1) DSAR (2) AODVand (3) EEABR
52 Simulation Analysis
521 Performance Metrics for Evaluating Routing ProtocolThe performance of routing protocol is evaluated by meansof end-to-end delay average throughput packet delivery raterouting overhead and so forth The statistical methods areintroduced as follows
(1) End-to-End Delay The average end-to-end delay is thetime required from the start of routing to the end of datatransmission We calculate it with the following formula
120591 = 1119873119873sum119894=1
(119905119886119894 minus 119905119887119894) (20)
where 120591 is the average end-to-end delay 119873 is the number ofsuccessful packet transmissions 119905119886119894 is the time that packet 119894arrives at the destination node and 119905119887119894 is the time packet 119894was generated
(2) Throughput Throughput is the maximum number ofpackets that a network successfully transmits per unit time
119879 = 1119879RE minus 119879RS
119873sum119894=1
119877119887 (119894) times 8 (21)
where119879 represents the throughput119877119887(119894) represents the num-ber of bytes of packet 119894 received successfully 119873 representsthe total number of packets received from the destination119879RE represents the reception time of the data packet and 119879RSrepresents the beginning of the data packet reception
(3) Packet Delivery Rate Packet delivery rate is the ratio ofthe total number of sending packets to the total number ofreceiving packets
(4) Routing Overhead
119873 = 119875119862119875119863 (22)
where119873 represents routing overhead 119875119862 represents the totalnumber of node send control packets and 119875119863 represents thetotal number of destination node receive data packets
522 The Network Performance Varies with the Packet SendRate of the Source Node The performance of the three algo-rithms varies with the average packet sending rate of thesource node in the network as shown in Figures 3ndash6
Figure 3 shows the relationship between the average end-to-end delay and the packet delivery rate of the source nodes
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
2 Wireless Communications and Mobile Computing
The paper is designed as follows The related work of thepaper is in Section 2 Section 3 proposes an ant colony routingalgorithm with reliability prediction under dual-channelconditions Section 4 proposes a route discovery processbased on joint optimization of dual-channel networks andSection 5 reveals simulation results and discussions FinallySection 6 concludes this paper
2 Related Work
Routing protocols designed for wireless ad hoc networksface many challenges especially in highly dynamic networksMany researchers have proposed algorithms to address this[6ndash13] These routing protocols can be divided into threecategories based on the different route discovery strategiesproactive on-demand and hybrid protocols In the activerouting protocols (eg DSDV [11] FSR [12] GSR [14] andHSR [6]) each network node periodically exchanges routinginformation with other nodes and each node must save therouting table whereas the proactive routing protocol hasa low delay it saves unnecessary routing information andconsumesmanynetwork resourcesWith on-demand routingprotocols (eg DSR [15] AODV [3] RDMAR [16] and LAR[17]) the routing information is created only when nodesare needed However there is a large delay with this type ofprotocol Both proactive and on-demand routing protocolscannot completely solve routing problems Therefore manyscholars have proposed hybrid routing protocols that com-bine the features of active and on-demand routing (eg ZRP[18] HWMP [19] and AOHR [20])
21 Representative Schemes for Comparison In the past fewyears swarm intelligence algorithms have become very popu-lar with ad hoc networks especially the ant colony optimiza-tion (ACO) algorithm The ACO algorithm is based on theantsrsquo foraging behavior in nature to find the optimal pathGraphically it has been applied widely in various fields [21]In the ant colony algorithm the intelligence of a single ant islimited but the ants can accomplish complex tasks via groupcooperation ACO has the same characteristics as the routingdesign of mobile ad hoc networks [22] G Dicaro proposedACO via AntNet [23] AntNet is based on the wired networkand the hop mechanism for data transmission When it istransplanted into ad hoc networks the power consumption ofsome nodes grows too high leading to unstable links Thusit is not suitable for mobile ad hoc networks
ARA protocol [24] proposed by Gunes et al in 2002 wasthe earliest application of ACO in ad hoc networks Howeverthe ARA protocol does not consider the energy balance ofnodes
Correia and Vazao proposed the SARA protocol [25]which was improved based on the ARA protocol SARA usesa restricted neighbor broadcast mechanism in which eachnode broadcasts the ant to all the neighbor nodes but onlyone neighbor is selected to send the forward ant SARA usesthe same routing maintenance and routing error handlingmechanism as the ARA protocol With routing maintenanceto reduce routing consumption the data packets also updatepheromones on the link During routing repair the depth
search algorithm is used to control the number of nodes torepair the routing and reduce the routing overhead caused bythe source node restart routing discovery process Comparedto theARAalgorithm the SARAalgorithmhas high through-put and low routing overhead but the route establishmentprocess of SARA takes a long time
AntHocNet [26] is a hybrid algorithm that combines theadvantages of AntNet and ARA protocols The routing pro-cess is carried out on demand and the routing mainte-nance process is carried out actively Compared with AODVthe packet delivery rate of AntHocNet protocol has beenimproved significantly However because of the active rout-ing maintenance mechanism the control overhead is muchhigher than AODV
The ACECR algorithm [27] was proposed by Zhou et alACECR is based on the ACO algorithm whose pheromoneupdates depend on two aspects the number of hops and theremaining energy However this algorithm does not considerstability and reliability
The R-ACO algorithm [28] preselects highly stable linksfor packet sending as far as possible to avoid the low linkstability of node processing packets Compared to LAR andAntHocNet R-ACO improves the success rate of data trans-mission while reducing communication overhead HoweverR-ACO algorithm increases the length of the path andrequires nodes to carry GPS devices
In [2] the level of pheromone is determined accordingto the routing length its congestion and the end-to-end pathreliabilityThis protocol provides high data delivery rates withlow end-to-end delay
For AOCR [29] the authors proposed an on-demand antcolony clustering routing protocol based on a weakly con-nected dominating set (AOCR) AOCR adopts the weaklyconnected dominating set as the auxiliary structure of cluster-ing nodesThe forward ants are only broadcast by the heads ofeach clusterThepheromone intensity depends on the averageresidual energy of the network and the minimum energyvalue of the nodes on the path AOCR adopts a pseudoran-dom proportional rule to select the most efficient routingCompared to other ant colony protocols AOCR requires lessstorage space and fewer network transmission resources toperform intelligent routing search
The authors in [30] proposed FTAR which avoids for-warding nodes with erroneous tendencies thus improvingthe performance of the network Compared with the DSRalgorithm E2FT [31] and AntHocNet FTAR improves datatransmission success rate and throughput However FTARuses an iterative method to obtain path confidence whichincreases the processing time of packets
The authors in [32] proposed an ant colony optimizationrouting algorithm (ie POSANT) based on geographic loca-tion information For the problem of the excessive numberof control packets and high transmission delays in ACOPOSANT combined the advantages of ACO and geographiclocation information Therefore it reduced the time of dis-covering routing and the number of ants generated by usinglocation information Compared to the GPSR algorithmPOSANT reduced the delay and increases the success rate of
Wireless Communications and Mobile Computing 3
data transmission but it did not consider the reliability of thepath
The authors in [33] proposed an enhanced congestioncontrol multipath routing method with ACO optimizationfor ad hoc networks which addressed the problem of linkblockage Additionally the load rapidly increases on the linkThey proposed an ACO-based multipath congestion controltechnique that varies the queue according to the load ina dynamic network Simulation showed that the proposedACO protocol had good performance However the algo-rithm did not use multiple channels and does not fundamen-tally solve the problem of network congestion
The authors in [5] presented a dual-channel network clus-tered routing protocol (DNCRP) a hybrid routing protocolDNCRP uses both 2-hop-distance neighbor cluster IDs andnode attribution information to search the path of interclusterrouting within the 2-hop-distance neighbor clusters perthe source cluster Simulation results show that DNCRP issuitable for high mobility networks
In [34] the energy of nodes was considered as a prior fac-tor for route choice However this scheme does not considerlink reliability as a factor for route choice and increase end-to-end delay
In summary there are two major limitations of the cur-rent routing algorithm First it does not use dual-channelstrategies Second there are very few routing schemes thatconsider the reliability of the integrated nodes and links as aprior factor for route-choosing
Hence there is a need to develop a unified routing proto-col which can fulfill the low end-to-end delay low routingoverhead and high reliability In this paper a reliable antcolony algorithm based on dual-channel conditions (DSAR)is proposed for ad hoc networks DSAR overcomes alllimitations of the previous schemes
3 An Ant Colony RoutingAlgorithm with Reliability Prediction underDual-Channel Conditions
31 Dual-Channel Joint Optimization Model In ad hoc net-works owing to the limited bandwidth of the nodes the datatransmission delay is high With technological developmenta node in ad hoc networks can be configured with twochannels which can reduce collisions increase bandwidthease network congestion and improve network performance[35] The dual-channel network communication model issimplified to the routing problem with a hierarchical mapwithout considering the channel assignment problem in thenetwork [36] In this paper we use a dual-channel layeredtransmission mode one channel as the control layer andanother as the data layer The control packets are transmittedin the control layer and the data packets are transmitted inthe data layer This double channeling eliminates messageconflict and reduces the delay of channel handoff If thecontrol layer is congested and the data layer has enoughbandwidth resources the dual-channel joint optimizationmode can transfer control packets in the control layer to thedata layer in real time to complete the joint scheduling of
the double-layer network and reduce congestion The dual-channel model is shown in Figure 1
32 Basic Ant Colony Algorithm for Ad Hoc Networks Whenan ant walks into an intersection it randomly selects a paththat has not been passed and releases pheromones Thesize of the pheromone is related to the path length Thelonger the path the smaller the pheromone When ants passby this intersection they will choose the path where thepheromone is large Thus a positive feedback is formedand the pheromone quantity on the optimal path is largerand the pheromone on other paths will become less overtime Simultaneously the whole ant colony can adapt tothe change of environment When ants suddenly encounterobstacles along the way they can quickly adjust their pathThus in the process of the entire colony finding the antsrsquopath a single antrsquos optimal path selection ability is limitedbut the ant colony has good self-organization because of theglobal pheromone Sharing path information the ants findthe optimal path via collective behavior of the ant communityThe ant colony algorithm has distributed parallel computercontrol which is easy to combine with other algorithms andhas strong robustness
The ant colony optimization algorithm has been success-fully applied to many optimization combinatorial problems[37] The ant foraging process is very similar to the routingproblem of ad hoc networks In this paper the nest and foodare compared to the source node and the destination nodein ad hoc networks The ant colony algorithm uses an antdecision table which comprises a node selection probabilityfrom a path and relevant local information The ants usethis decision table to guide their search of the mobile spacein the optimal region which is the process of forming therouting tableThus the ant colony algorithmcan be used in adhoc networks Through the pheromone mechanism the antssearch for and maintain optimal routing The mechanism ofevaporation updates the pheromone of each node which canquickly adapt to the needs of the dynamic changes of ad hocnetworks
In these networks the network topology model is thewireless graph 119866(119881 119864) where 119881 is a network node and 119864 isthe link between two nodes At time 119905 there are 119886119894(119905) antsThetotal number of ants in the network is119898 = sum119899119894=1 119886119894(119905) 119875119894119895(119905) isthe probability of choosing link 119864119894119895 for ant119870 at time 119905
119875119896119894119895 (119905) =
[120591119894119895 (119905)]120572 [120578119894119895]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895]120573
119895 isin allowed119896
0 else
(1)
where 120591119894119895(119905) is the strength of the pheromone in the link 119864119894119895120572 is a parameter to measure the trajectory of pheromones120578119894119895 is visibility between node 119894 and node 119895 which is generallydefined as 1119889119894119895 (119889119894119895 is the distance between node 119894 and node119895) 120573 is a parameter that measures visibility and allowed119896 is acollection of nodes that have not been visited
4 Wireless Communications and Mobile Computing
LL
InterfaceQueue
MAC
Networkinterface
LL
InterfaceQueue
MAC
ARP ARP
Application
RoutingAgent
Channel 1
Networkinterface
Channel 2
PropagationMode
Addr
ess
Mul
tiple
xer Po
rtM
ultip
lexe
r
Figure 1 Dual-channel joint optimization mode
The pheromone update formula on each path in ad hocnetworks is as follows
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) + Δ120591119894119895 (2)
where 120588 is the pheromone volatilization coefficient which isa constant between 0 and 1 and Δ120591119894119895 is the increment of thepheromone of ants passing through links 119894 and 119895
Δ120591119894119895 = 119897sum119896=1
Δ120591119896119894119895 (3)
33 Route Reliability In ad hoc networks there are twomainreasons for path breaking One is the movement of nodeson the communication path and the other is the nodeswithdrawing from the network because of energy depletionThus we select relatively reliable nodes and links Then thepath stability (PS) factor is introduced to judge the stability ofthe path During the establishment of QoS routing the pathwith the strongest stability is selected from themultiple pathssatisfying the QoS requirements reducing the probabilityof path breaking In this paper the path stability factor isthe function of the link stability factor and the node energystability factor
331TheNode Energy Stability Factor Suppose that there are119895 intermediate nodes in the path 119875119894 119873119894 = 1198991198941 1198991198942 119899119894119895
In this paper we define the node energy stability fac-tor
ES119894119895 = 119864currrent (119894119895)119864initial (119894119895) (4)
where 119864initial(119894119895) is the initial energy of node 119895 in path 119875119894 and119864currrent(119894119895) is the current remaining energy of node 119895 in path119875119894Because a node in the path cannot be used owing to the
exhausted energy the energy stability factor of the path is theminimum energy stability factor of all nodes in the path 119875119894
ES119894 = min ES119894119895 119897119894119895 isin 119901119894 (5)
332 The Link Stability Factor In this paper we define thelink reliability factor as the remaining lifetime of the linkThe communication radius of each node is 119877 Each nodeis equipped with GPS thus every node can perceive thelocation speed time of nodes and period and send itsown coordinates and speed information to neighbor nodesAccording to the location information of nodes the remain-ing lifetime of each link can be predicted and the link stabilityfactor can be obtained As shown in Figure 2 the initialdistance between the two nodes119872 and119873 is 119889
The relative motion of the two nodes is equivalent toone node moving while the other node is stationary Thecoordinates of node119873 relative to the stationary node119872 are
Wireless Communications and Mobile Computing 5
M
N
R
MN
Figure 2 Calculate the lifetime of the link119872119873
(119909119873 minus 119909119872 119910119873 minus 119910119872) Thus the distance between119872 and119873 is119889 = radic1199091198721198732 + 1199101198721198732 According to the relative movement ofthe two nodes after time 119905 node119872 relative to position119873 is(1199091015840119872119873 1199101015840119872119873)
1199091015840119872119873 = 119909119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 cos 120579119872119873
1199101015840119872119873 = 119910119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 sin 120579119872119873 (6)
When the distance between 119872 and 119873 is 119877 the link between119872 and119873 is broken
1198891015840 = radic(1199091015840119872119873)2 + (1199101015840119872119873)2 = 119877 (7)
Take (6) into (7)
10038161003816100381610038161003816997888997888997888V119873119872100381610038161003816100381610038162 1199052 + 2 10038161003816100381610038161003816997888997888997888V119873119872
10038161003816100381610038161003816 (119909119873119872 cos 120579119873119872 + 119910119873119872 sin 120579119873119872) 119905+ 1198892 minus 1198772 = 0 (8)
where997888V119872 = (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894 + (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 997888119895 997888V119873 = (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873) 997888119894 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873) 997888119895 997888997888997888V119873119872 = 997888V119873 minus 997888V119872= (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894
+ (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 99788811989510038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic(10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)2 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872)2120579119873119872 = tanminus1(
10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 12057911987210038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)
(9)
Because 119905 cannot be negative 119905 becomes the following
119905 = radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)210038161003816100381610038161003816997888997888997888rarrV11987311987210038161003816100381610038161003816
minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
(10)
Link stability is
LS119872119873 = 119905119905max= 119905
119877 10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)2119877minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)119877
(11)
There is a 119896 link in a path between the source node 119878 and thedestination node 119863 The link stability of the path dependson the minimum link stability factor in the 119896-segment linkThus we define the link reliability factor of path 119875119894 as theminimum link stability factor
LS119894 = min LS119872119873 119897119872119873 isin 119901119894 (12)
333 The Path Reliability Factor Considering the link stabil-ity factor and node energy stability factor the comprehensivereliability factor of the path is defined as
PS119894 = 11ES119894 + 1LS119894 =ES119894 times LS119894ES119894 + LS119894
(13)
Whether the communication path between the source nodeand the destination node is stable depends on the worst linksand nodes in the path because the entire communication isinterrupted when there is a broken link or when a node exitsthe network because of exhausted battery energy After usingthe abovemethod if either of the two stability factors is smallthe value of PS119894 will be small Thus path reliability will bepoor When the source node finds multiple paths satisfyingthe QoS requirements the one with the largest PS is selectedThus PS = max PS119894 119875119894 isin 1198754 Route Discovery Process Based on JointOptimization of Dual-Channel Networks
To improve channel utilization based on the double-channelmodel and ant colony optimization algorithm the two-interlayer joint optimization routing mode is proposed Itcan increase bandwidth and make full use of idle resourcesbetween different layers In Mode 1 the routing service in thecontrol layer can only be transmitted in the control layer Ifthe control layer does not have enough channel resourcesthe service will be rejected To reduce the blocking rate of thecontrol layer joint optimization Mode 2 is proposed Whenthe data layer has enough idle resources the control packetsin the control layer can be transmitted to the data layer inreal time to realize the joint optimization of the two-layernetwork The specific routing process of the two modes is asfollows
Mode 1 It is the route discovery mode on the control layer ofDSAR
(1) Broadcast a Hello Message At the control layer eachnetwork node periodically sends a 1-hop hello message The
6 Wireless Communications and Mobile Computing
Table 1 The format of the hello message
Type Src addr Scr 119883 Scr 119884 Scr 119881 119873 energy
Table 2 The structure of neighbor
119873 addr 119873 energy LS Hops119873 119883 Nr 119884 119873 119881 Phenomenon
format of the hello message is shown in Table 1 whereldquoTyperdquo denotes packet type ldquoSrc addrrdquo is the address of thesource node that sends the hello message ldquoScr Xrdquo are the Xcoordinates of node ldquoScr Yrdquo are the Y coordinates of nodeldquoScr Vrdquo is velocity vector of node and ldquoN energyrdquo is residualenergy of node
(2) Establish the Neighbor List Each node establishes itsneighbor list by receiving hello messages sent by neighbornodes in real time In this paper each network node hasa GPS positioning device to obtain its geographic locationinformation Each network node sends hello packets period-ically and accepts those sent by neighbor nodes Each hellopacket contains node coordinates and speed information andthe neighbor list of each node The neighbor list contains thelocation vector information of all neighboring nodes Thuseach network node can obtain the location information ofneighbor nodesThe frequency of packet transmission can beset according to differentmotion scenarios and the transmis-sion frequency of hello messages can be higher for scenarioswhere network topology changes rapidly In this paper wedesign hellomessaging with a sending interval of 1 s If a nodedoes not receive neighbor hello packets in 1 s the link of thisnode is broken and the node deletes it from its neighbortable The node updates its neighbor list immediately afterthe hello message is received Otherwise the comprehensivereliability of the node is calculated according to formula (13)The format of neighbor message is shown in Table 2 whereldquo119873 addrrdquo is the address of the source node that sends thehello message ldquo119873 energyrdquo is residual energy of neighbornode ldquo119873 119883rdquo are the119883 coordinates of neighbor node ldquo119873 119884rdquoare the 119884 coordinates of neighbor node 119873 119881rdquo is velocityvector of neighbor node ldquoLSrdquo is the stability of neighbornode ldquoHopsrdquo denotes total hops of a route passing throughthis neighbor node and ldquoPhenomenonrdquo denotes the value ofpheromone
(3) Send a Forward Ant and Update Pheromone When thesource node 119878 has data packets sent to the destinationnode119863 the source node 119878 looks at the routing informationtable namely the pheromone table If there is no routinginformation and the control layer has enough bandwidththe forward ant is broadcast in the control layer If there isa route to node119863 the packet is sent directly to the data layerIn the initial stage of routing establishment the source nodebroadcasts a certain number of forward ants at the controllayer The unicast or broadcast of each intermediate nodebetween source node 119878 and destination node 119863 depends onwhether the intermediate node has pheromone If there is
Table 3 The format of forward ant message
Type Fant addr Fant Seqno TTLScr 119884 Scr 119881 119864sum Fd119896
Seqno 0Seqno 1Seqno 119899
pheromone at each intermediate node the probability thatant 119870 selects the next hop neighbor node 119895 is calculatedaccording to the following equation
119875119896119894119895 (119905)
=
[120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573
119895 isin allowed119896
0 else(14)
where 119875119896119894119895(119905) is the probability that ant 119870 selects the next hopneighbor node 119895 when node 119894moves toward the destinationnode 120591119894119895(119905) is the size of the pheromone of node 119894 at time119905 in link 119894119895 120578119894119895(119905) is a visual function of node 119894 at time 119905from node 119894 to node 119895 allowed119896 is a collection of neighbornodes of node 119894 120572 and 120573 are adjustment coefficients 120572 is arelative importance coefficient of the residual pheromone 120573is a relative importance coefficient of heuristic informationThe pheromone of node is updated as equation
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) (15)
The format of the forward ant message is shown inTable 3 where ldquoTyperdquo denotes the packet type ldquoFant addrrdquois the address of source node ldquoFant Seqnordquo is the sequencenumber of forward ants generated by source nodes ldquoSeqnordquodenotes node address visited by the forward ant ldquoTTLrdquo is thesurvival time of forward ant and ldquoFd119896rdquo denotes the numberof hops that the forward ant moves to the current node
(4) Send a Backward Ant and Update Pheromone When aforward ant arrives at the destination node it turns into abackward ant The backward ant returns to the source nodealong its former path If a fault occurs in the link of thenext hop caused by the movement of a node in the path tothe source node then the backward ant will be discardedWhen the backward ant returns to node 119894 from node 119895 thepheromone of node 119894 is updated according to the pheromoneupdate of (16)
120591119894119895 (119905 + 1) = 120591119894119895 (119905)1 minus 120588 + Δ120591119894119895 (119905) (16)
Δ120591119894119895 (119905) = 119898sum119896=1
Δ120591119896119894119895 (119905) (17)
Wireless Communications and Mobile Computing 7
Table 4 The format of backward ant message
Type Visitednode 119864min 119864avg Bd119896
Δ120591119896119894119895 (119905) = PS119894 (18)
120578119896119894119895 (119905) = 1Fd119896
(19)
The format of the backward ant message is shown inTable 4 where ldquoTyperdquo denotes the packet type ldquoVisitednoderdquodenotes the ID of node visited ldquo119864minrdquo is the energy value ofthe minimum energy node on the path that the backward antpasses through ldquo119864avgrdquo is the residual average energy of theant 119870 to the current node and ldquoBd119896rdquo denotes the number ofhops experienced by the backward ant 119870 to node 119895Mode 2 It is the route discovery process based on jointoptimization of dual-channel networks
When the control layer is congested joint optimizationMode 2 is proposed That is when the data layer has enoughidle resources the control packets in the control layer can betransmitted to the data layer in real time to realize the jointoptimization of the two-layer network The specific routingprocess of Mode 2 is as follows
(1) If the control layer has enough network resourcesthe forward ant is routed via Mode 1 in the control layerOtherwise turn to (2)
(2) When the forward ant119870 moves to node 119894 in the datatransport layer the forward ant 119870 looks at whether thereare available channel resources for ants to find paths withneighboring nodes of 119894 If not ant 119870 stops and refuses toperform the routing lookup service Otherwise it turns to (3)
(3) Ant 119870 performs the routing service in the data trans-mission layer and finds the next hop node 119895 in Mode 1
(4)When reachingnode 119895 ant119870first investigateswhetherthere are enough channel resources between node 119895 and itsneighbor nodes in the control layer to perform the routingservice If not ant 119870 continues to perform the path-findingservice in the data transport layer Otherwise ant 119870 returnsto the control layer and searches the optimal path of serviceaccording to Mode 1
5 Simulation and Analysis
In this paper to verify the reliability of the DSAR protocolNS-2 is selected and the DSAR algorithm is compared tothe EEABR algorithm [34] and the AODV algorithm AODVis a classic routing algorithm and EEABR is a successfulapplication of the ant colony algorithm in wireless ad hocnetworks
51 Simulation Setup In a wireless simulation environmenteach mobile network node is randomly distributed in the1000m times 1000m area 50 nodes are randomly arrangedaccording to the random way-point model The communi-cation radius of each node is 250m MAC layer adopts dual-channel mode The packet length is 512 b and the send rate
varies from 1 to 16 packetss The evaporation of pheromoneoccurs every 1 s The evaporation rate 120588 is set to 02 Eachvalue of 120572 and 120573 is set to 20 and 15 respectively Thesimulation time is 1000 s To reduce random errors theexperimental results will be the average of the 10 experimentsSimulation algorithm routing layers are (1) DSAR (2) AODVand (3) EEABR
52 Simulation Analysis
521 Performance Metrics for Evaluating Routing ProtocolThe performance of routing protocol is evaluated by meansof end-to-end delay average throughput packet delivery raterouting overhead and so forth The statistical methods areintroduced as follows
(1) End-to-End Delay The average end-to-end delay is thetime required from the start of routing to the end of datatransmission We calculate it with the following formula
120591 = 1119873119873sum119894=1
(119905119886119894 minus 119905119887119894) (20)
where 120591 is the average end-to-end delay 119873 is the number ofsuccessful packet transmissions 119905119886119894 is the time that packet 119894arrives at the destination node and 119905119887119894 is the time packet 119894was generated
(2) Throughput Throughput is the maximum number ofpackets that a network successfully transmits per unit time
119879 = 1119879RE minus 119879RS
119873sum119894=1
119877119887 (119894) times 8 (21)
where119879 represents the throughput119877119887(119894) represents the num-ber of bytes of packet 119894 received successfully 119873 representsthe total number of packets received from the destination119879RE represents the reception time of the data packet and 119879RSrepresents the beginning of the data packet reception
(3) Packet Delivery Rate Packet delivery rate is the ratio ofthe total number of sending packets to the total number ofreceiving packets
(4) Routing Overhead
119873 = 119875119862119875119863 (22)
where119873 represents routing overhead 119875119862 represents the totalnumber of node send control packets and 119875119863 represents thetotal number of destination node receive data packets
522 The Network Performance Varies with the Packet SendRate of the Source Node The performance of the three algo-rithms varies with the average packet sending rate of thesource node in the network as shown in Figures 3ndash6
Figure 3 shows the relationship between the average end-to-end delay and the packet delivery rate of the source nodes
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Wireless Communications and Mobile Computing 3
data transmission but it did not consider the reliability of thepath
The authors in [33] proposed an enhanced congestioncontrol multipath routing method with ACO optimizationfor ad hoc networks which addressed the problem of linkblockage Additionally the load rapidly increases on the linkThey proposed an ACO-based multipath congestion controltechnique that varies the queue according to the load ina dynamic network Simulation showed that the proposedACO protocol had good performance However the algo-rithm did not use multiple channels and does not fundamen-tally solve the problem of network congestion
The authors in [5] presented a dual-channel network clus-tered routing protocol (DNCRP) a hybrid routing protocolDNCRP uses both 2-hop-distance neighbor cluster IDs andnode attribution information to search the path of interclusterrouting within the 2-hop-distance neighbor clusters perthe source cluster Simulation results show that DNCRP issuitable for high mobility networks
In [34] the energy of nodes was considered as a prior fac-tor for route choice However this scheme does not considerlink reliability as a factor for route choice and increase end-to-end delay
In summary there are two major limitations of the cur-rent routing algorithm First it does not use dual-channelstrategies Second there are very few routing schemes thatconsider the reliability of the integrated nodes and links as aprior factor for route-choosing
Hence there is a need to develop a unified routing proto-col which can fulfill the low end-to-end delay low routingoverhead and high reliability In this paper a reliable antcolony algorithm based on dual-channel conditions (DSAR)is proposed for ad hoc networks DSAR overcomes alllimitations of the previous schemes
3 An Ant Colony RoutingAlgorithm with Reliability Prediction underDual-Channel Conditions
31 Dual-Channel Joint Optimization Model In ad hoc net-works owing to the limited bandwidth of the nodes the datatransmission delay is high With technological developmenta node in ad hoc networks can be configured with twochannels which can reduce collisions increase bandwidthease network congestion and improve network performance[35] The dual-channel network communication model issimplified to the routing problem with a hierarchical mapwithout considering the channel assignment problem in thenetwork [36] In this paper we use a dual-channel layeredtransmission mode one channel as the control layer andanother as the data layer The control packets are transmittedin the control layer and the data packets are transmitted inthe data layer This double channeling eliminates messageconflict and reduces the delay of channel handoff If thecontrol layer is congested and the data layer has enoughbandwidth resources the dual-channel joint optimizationmode can transfer control packets in the control layer to thedata layer in real time to complete the joint scheduling of
the double-layer network and reduce congestion The dual-channel model is shown in Figure 1
32 Basic Ant Colony Algorithm for Ad Hoc Networks Whenan ant walks into an intersection it randomly selects a paththat has not been passed and releases pheromones Thesize of the pheromone is related to the path length Thelonger the path the smaller the pheromone When ants passby this intersection they will choose the path where thepheromone is large Thus a positive feedback is formedand the pheromone quantity on the optimal path is largerand the pheromone on other paths will become less overtime Simultaneously the whole ant colony can adapt tothe change of environment When ants suddenly encounterobstacles along the way they can quickly adjust their pathThus in the process of the entire colony finding the antsrsquopath a single antrsquos optimal path selection ability is limitedbut the ant colony has good self-organization because of theglobal pheromone Sharing path information the ants findthe optimal path via collective behavior of the ant communityThe ant colony algorithm has distributed parallel computercontrol which is easy to combine with other algorithms andhas strong robustness
The ant colony optimization algorithm has been success-fully applied to many optimization combinatorial problems[37] The ant foraging process is very similar to the routingproblem of ad hoc networks In this paper the nest and foodare compared to the source node and the destination nodein ad hoc networks The ant colony algorithm uses an antdecision table which comprises a node selection probabilityfrom a path and relevant local information The ants usethis decision table to guide their search of the mobile spacein the optimal region which is the process of forming therouting tableThus the ant colony algorithmcan be used in adhoc networks Through the pheromone mechanism the antssearch for and maintain optimal routing The mechanism ofevaporation updates the pheromone of each node which canquickly adapt to the needs of the dynamic changes of ad hocnetworks
In these networks the network topology model is thewireless graph 119866(119881 119864) where 119881 is a network node and 119864 isthe link between two nodes At time 119905 there are 119886119894(119905) antsThetotal number of ants in the network is119898 = sum119899119894=1 119886119894(119905) 119875119894119895(119905) isthe probability of choosing link 119864119894119895 for ant119870 at time 119905
119875119896119894119895 (119905) =
[120591119894119895 (119905)]120572 [120578119894119895]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895]120573
119895 isin allowed119896
0 else
(1)
where 120591119894119895(119905) is the strength of the pheromone in the link 119864119894119895120572 is a parameter to measure the trajectory of pheromones120578119894119895 is visibility between node 119894 and node 119895 which is generallydefined as 1119889119894119895 (119889119894119895 is the distance between node 119894 and node119895) 120573 is a parameter that measures visibility and allowed119896 is acollection of nodes that have not been visited
4 Wireless Communications and Mobile Computing
LL
InterfaceQueue
MAC
Networkinterface
LL
InterfaceQueue
MAC
ARP ARP
Application
RoutingAgent
Channel 1
Networkinterface
Channel 2
PropagationMode
Addr
ess
Mul
tiple
xer Po
rtM
ultip
lexe
r
Figure 1 Dual-channel joint optimization mode
The pheromone update formula on each path in ad hocnetworks is as follows
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) + Δ120591119894119895 (2)
where 120588 is the pheromone volatilization coefficient which isa constant between 0 and 1 and Δ120591119894119895 is the increment of thepheromone of ants passing through links 119894 and 119895
Δ120591119894119895 = 119897sum119896=1
Δ120591119896119894119895 (3)
33 Route Reliability In ad hoc networks there are twomainreasons for path breaking One is the movement of nodeson the communication path and the other is the nodeswithdrawing from the network because of energy depletionThus we select relatively reliable nodes and links Then thepath stability (PS) factor is introduced to judge the stability ofthe path During the establishment of QoS routing the pathwith the strongest stability is selected from themultiple pathssatisfying the QoS requirements reducing the probabilityof path breaking In this paper the path stability factor isthe function of the link stability factor and the node energystability factor
331TheNode Energy Stability Factor Suppose that there are119895 intermediate nodes in the path 119875119894 119873119894 = 1198991198941 1198991198942 119899119894119895
In this paper we define the node energy stability fac-tor
ES119894119895 = 119864currrent (119894119895)119864initial (119894119895) (4)
where 119864initial(119894119895) is the initial energy of node 119895 in path 119875119894 and119864currrent(119894119895) is the current remaining energy of node 119895 in path119875119894Because a node in the path cannot be used owing to the
exhausted energy the energy stability factor of the path is theminimum energy stability factor of all nodes in the path 119875119894
ES119894 = min ES119894119895 119897119894119895 isin 119901119894 (5)
332 The Link Stability Factor In this paper we define thelink reliability factor as the remaining lifetime of the linkThe communication radius of each node is 119877 Each nodeis equipped with GPS thus every node can perceive thelocation speed time of nodes and period and send itsown coordinates and speed information to neighbor nodesAccording to the location information of nodes the remain-ing lifetime of each link can be predicted and the link stabilityfactor can be obtained As shown in Figure 2 the initialdistance between the two nodes119872 and119873 is 119889
The relative motion of the two nodes is equivalent toone node moving while the other node is stationary Thecoordinates of node119873 relative to the stationary node119872 are
Wireless Communications and Mobile Computing 5
M
N
R
MN
Figure 2 Calculate the lifetime of the link119872119873
(119909119873 minus 119909119872 119910119873 minus 119910119872) Thus the distance between119872 and119873 is119889 = radic1199091198721198732 + 1199101198721198732 According to the relative movement ofthe two nodes after time 119905 node119872 relative to position119873 is(1199091015840119872119873 1199101015840119872119873)
1199091015840119872119873 = 119909119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 cos 120579119872119873
1199101015840119872119873 = 119910119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 sin 120579119872119873 (6)
When the distance between 119872 and 119873 is 119877 the link between119872 and119873 is broken
1198891015840 = radic(1199091015840119872119873)2 + (1199101015840119872119873)2 = 119877 (7)
Take (6) into (7)
10038161003816100381610038161003816997888997888997888V119873119872100381610038161003816100381610038162 1199052 + 2 10038161003816100381610038161003816997888997888997888V119873119872
10038161003816100381610038161003816 (119909119873119872 cos 120579119873119872 + 119910119873119872 sin 120579119873119872) 119905+ 1198892 minus 1198772 = 0 (8)
where997888V119872 = (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894 + (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 997888119895 997888V119873 = (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873) 997888119894 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873) 997888119895 997888997888997888V119873119872 = 997888V119873 minus 997888V119872= (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894
+ (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 99788811989510038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic(10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)2 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872)2120579119873119872 = tanminus1(
10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 12057911987210038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)
(9)
Because 119905 cannot be negative 119905 becomes the following
119905 = radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)210038161003816100381610038161003816997888997888997888rarrV11987311987210038161003816100381610038161003816
minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
(10)
Link stability is
LS119872119873 = 119905119905max= 119905
119877 10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)2119877minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)119877
(11)
There is a 119896 link in a path between the source node 119878 and thedestination node 119863 The link stability of the path dependson the minimum link stability factor in the 119896-segment linkThus we define the link reliability factor of path 119875119894 as theminimum link stability factor
LS119894 = min LS119872119873 119897119872119873 isin 119901119894 (12)
333 The Path Reliability Factor Considering the link stabil-ity factor and node energy stability factor the comprehensivereliability factor of the path is defined as
PS119894 = 11ES119894 + 1LS119894 =ES119894 times LS119894ES119894 + LS119894
(13)
Whether the communication path between the source nodeand the destination node is stable depends on the worst linksand nodes in the path because the entire communication isinterrupted when there is a broken link or when a node exitsthe network because of exhausted battery energy After usingthe abovemethod if either of the two stability factors is smallthe value of PS119894 will be small Thus path reliability will bepoor When the source node finds multiple paths satisfyingthe QoS requirements the one with the largest PS is selectedThus PS = max PS119894 119875119894 isin 1198754 Route Discovery Process Based on JointOptimization of Dual-Channel Networks
To improve channel utilization based on the double-channelmodel and ant colony optimization algorithm the two-interlayer joint optimization routing mode is proposed Itcan increase bandwidth and make full use of idle resourcesbetween different layers In Mode 1 the routing service in thecontrol layer can only be transmitted in the control layer Ifthe control layer does not have enough channel resourcesthe service will be rejected To reduce the blocking rate of thecontrol layer joint optimization Mode 2 is proposed Whenthe data layer has enough idle resources the control packetsin the control layer can be transmitted to the data layer inreal time to realize the joint optimization of the two-layernetwork The specific routing process of the two modes is asfollows
Mode 1 It is the route discovery mode on the control layer ofDSAR
(1) Broadcast a Hello Message At the control layer eachnetwork node periodically sends a 1-hop hello message The
6 Wireless Communications and Mobile Computing
Table 1 The format of the hello message
Type Src addr Scr 119883 Scr 119884 Scr 119881 119873 energy
Table 2 The structure of neighbor
119873 addr 119873 energy LS Hops119873 119883 Nr 119884 119873 119881 Phenomenon
format of the hello message is shown in Table 1 whereldquoTyperdquo denotes packet type ldquoSrc addrrdquo is the address of thesource node that sends the hello message ldquoScr Xrdquo are the Xcoordinates of node ldquoScr Yrdquo are the Y coordinates of nodeldquoScr Vrdquo is velocity vector of node and ldquoN energyrdquo is residualenergy of node
(2) Establish the Neighbor List Each node establishes itsneighbor list by receiving hello messages sent by neighbornodes in real time In this paper each network node hasa GPS positioning device to obtain its geographic locationinformation Each network node sends hello packets period-ically and accepts those sent by neighbor nodes Each hellopacket contains node coordinates and speed information andthe neighbor list of each node The neighbor list contains thelocation vector information of all neighboring nodes Thuseach network node can obtain the location information ofneighbor nodesThe frequency of packet transmission can beset according to differentmotion scenarios and the transmis-sion frequency of hello messages can be higher for scenarioswhere network topology changes rapidly In this paper wedesign hellomessaging with a sending interval of 1 s If a nodedoes not receive neighbor hello packets in 1 s the link of thisnode is broken and the node deletes it from its neighbortable The node updates its neighbor list immediately afterthe hello message is received Otherwise the comprehensivereliability of the node is calculated according to formula (13)The format of neighbor message is shown in Table 2 whereldquo119873 addrrdquo is the address of the source node that sends thehello message ldquo119873 energyrdquo is residual energy of neighbornode ldquo119873 119883rdquo are the119883 coordinates of neighbor node ldquo119873 119884rdquoare the 119884 coordinates of neighbor node 119873 119881rdquo is velocityvector of neighbor node ldquoLSrdquo is the stability of neighbornode ldquoHopsrdquo denotes total hops of a route passing throughthis neighbor node and ldquoPhenomenonrdquo denotes the value ofpheromone
(3) Send a Forward Ant and Update Pheromone When thesource node 119878 has data packets sent to the destinationnode119863 the source node 119878 looks at the routing informationtable namely the pheromone table If there is no routinginformation and the control layer has enough bandwidththe forward ant is broadcast in the control layer If there isa route to node119863 the packet is sent directly to the data layerIn the initial stage of routing establishment the source nodebroadcasts a certain number of forward ants at the controllayer The unicast or broadcast of each intermediate nodebetween source node 119878 and destination node 119863 depends onwhether the intermediate node has pheromone If there is
Table 3 The format of forward ant message
Type Fant addr Fant Seqno TTLScr 119884 Scr 119881 119864sum Fd119896
Seqno 0Seqno 1Seqno 119899
pheromone at each intermediate node the probability thatant 119870 selects the next hop neighbor node 119895 is calculatedaccording to the following equation
119875119896119894119895 (119905)
=
[120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573
119895 isin allowed119896
0 else(14)
where 119875119896119894119895(119905) is the probability that ant 119870 selects the next hopneighbor node 119895 when node 119894moves toward the destinationnode 120591119894119895(119905) is the size of the pheromone of node 119894 at time119905 in link 119894119895 120578119894119895(119905) is a visual function of node 119894 at time 119905from node 119894 to node 119895 allowed119896 is a collection of neighbornodes of node 119894 120572 and 120573 are adjustment coefficients 120572 is arelative importance coefficient of the residual pheromone 120573is a relative importance coefficient of heuristic informationThe pheromone of node is updated as equation
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) (15)
The format of the forward ant message is shown inTable 3 where ldquoTyperdquo denotes the packet type ldquoFant addrrdquois the address of source node ldquoFant Seqnordquo is the sequencenumber of forward ants generated by source nodes ldquoSeqnordquodenotes node address visited by the forward ant ldquoTTLrdquo is thesurvival time of forward ant and ldquoFd119896rdquo denotes the numberof hops that the forward ant moves to the current node
(4) Send a Backward Ant and Update Pheromone When aforward ant arrives at the destination node it turns into abackward ant The backward ant returns to the source nodealong its former path If a fault occurs in the link of thenext hop caused by the movement of a node in the path tothe source node then the backward ant will be discardedWhen the backward ant returns to node 119894 from node 119895 thepheromone of node 119894 is updated according to the pheromoneupdate of (16)
120591119894119895 (119905 + 1) = 120591119894119895 (119905)1 minus 120588 + Δ120591119894119895 (119905) (16)
Δ120591119894119895 (119905) = 119898sum119896=1
Δ120591119896119894119895 (119905) (17)
Wireless Communications and Mobile Computing 7
Table 4 The format of backward ant message
Type Visitednode 119864min 119864avg Bd119896
Δ120591119896119894119895 (119905) = PS119894 (18)
120578119896119894119895 (119905) = 1Fd119896
(19)
The format of the backward ant message is shown inTable 4 where ldquoTyperdquo denotes the packet type ldquoVisitednoderdquodenotes the ID of node visited ldquo119864minrdquo is the energy value ofthe minimum energy node on the path that the backward antpasses through ldquo119864avgrdquo is the residual average energy of theant 119870 to the current node and ldquoBd119896rdquo denotes the number ofhops experienced by the backward ant 119870 to node 119895Mode 2 It is the route discovery process based on jointoptimization of dual-channel networks
When the control layer is congested joint optimizationMode 2 is proposed That is when the data layer has enoughidle resources the control packets in the control layer can betransmitted to the data layer in real time to realize the jointoptimization of the two-layer network The specific routingprocess of Mode 2 is as follows
(1) If the control layer has enough network resourcesthe forward ant is routed via Mode 1 in the control layerOtherwise turn to (2)
(2) When the forward ant119870 moves to node 119894 in the datatransport layer the forward ant 119870 looks at whether thereare available channel resources for ants to find paths withneighboring nodes of 119894 If not ant 119870 stops and refuses toperform the routing lookup service Otherwise it turns to (3)
(3) Ant 119870 performs the routing service in the data trans-mission layer and finds the next hop node 119895 in Mode 1
(4)When reachingnode 119895 ant119870first investigateswhetherthere are enough channel resources between node 119895 and itsneighbor nodes in the control layer to perform the routingservice If not ant 119870 continues to perform the path-findingservice in the data transport layer Otherwise ant 119870 returnsto the control layer and searches the optimal path of serviceaccording to Mode 1
5 Simulation and Analysis
In this paper to verify the reliability of the DSAR protocolNS-2 is selected and the DSAR algorithm is compared tothe EEABR algorithm [34] and the AODV algorithm AODVis a classic routing algorithm and EEABR is a successfulapplication of the ant colony algorithm in wireless ad hocnetworks
51 Simulation Setup In a wireless simulation environmenteach mobile network node is randomly distributed in the1000m times 1000m area 50 nodes are randomly arrangedaccording to the random way-point model The communi-cation radius of each node is 250m MAC layer adopts dual-channel mode The packet length is 512 b and the send rate
varies from 1 to 16 packetss The evaporation of pheromoneoccurs every 1 s The evaporation rate 120588 is set to 02 Eachvalue of 120572 and 120573 is set to 20 and 15 respectively Thesimulation time is 1000 s To reduce random errors theexperimental results will be the average of the 10 experimentsSimulation algorithm routing layers are (1) DSAR (2) AODVand (3) EEABR
52 Simulation Analysis
521 Performance Metrics for Evaluating Routing ProtocolThe performance of routing protocol is evaluated by meansof end-to-end delay average throughput packet delivery raterouting overhead and so forth The statistical methods areintroduced as follows
(1) End-to-End Delay The average end-to-end delay is thetime required from the start of routing to the end of datatransmission We calculate it with the following formula
120591 = 1119873119873sum119894=1
(119905119886119894 minus 119905119887119894) (20)
where 120591 is the average end-to-end delay 119873 is the number ofsuccessful packet transmissions 119905119886119894 is the time that packet 119894arrives at the destination node and 119905119887119894 is the time packet 119894was generated
(2) Throughput Throughput is the maximum number ofpackets that a network successfully transmits per unit time
119879 = 1119879RE minus 119879RS
119873sum119894=1
119877119887 (119894) times 8 (21)
where119879 represents the throughput119877119887(119894) represents the num-ber of bytes of packet 119894 received successfully 119873 representsthe total number of packets received from the destination119879RE represents the reception time of the data packet and 119879RSrepresents the beginning of the data packet reception
(3) Packet Delivery Rate Packet delivery rate is the ratio ofthe total number of sending packets to the total number ofreceiving packets
(4) Routing Overhead
119873 = 119875119862119875119863 (22)
where119873 represents routing overhead 119875119862 represents the totalnumber of node send control packets and 119875119863 represents thetotal number of destination node receive data packets
522 The Network Performance Varies with the Packet SendRate of the Source Node The performance of the three algo-rithms varies with the average packet sending rate of thesource node in the network as shown in Figures 3ndash6
Figure 3 shows the relationship between the average end-to-end delay and the packet delivery rate of the source nodes
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
4 Wireless Communications and Mobile Computing
LL
InterfaceQueue
MAC
Networkinterface
LL
InterfaceQueue
MAC
ARP ARP
Application
RoutingAgent
Channel 1
Networkinterface
Channel 2
PropagationMode
Addr
ess
Mul
tiple
xer Po
rtM
ultip
lexe
r
Figure 1 Dual-channel joint optimization mode
The pheromone update formula on each path in ad hocnetworks is as follows
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) + Δ120591119894119895 (2)
where 120588 is the pheromone volatilization coefficient which isa constant between 0 and 1 and Δ120591119894119895 is the increment of thepheromone of ants passing through links 119894 and 119895
Δ120591119894119895 = 119897sum119896=1
Δ120591119896119894119895 (3)
33 Route Reliability In ad hoc networks there are twomainreasons for path breaking One is the movement of nodeson the communication path and the other is the nodeswithdrawing from the network because of energy depletionThus we select relatively reliable nodes and links Then thepath stability (PS) factor is introduced to judge the stability ofthe path During the establishment of QoS routing the pathwith the strongest stability is selected from themultiple pathssatisfying the QoS requirements reducing the probabilityof path breaking In this paper the path stability factor isthe function of the link stability factor and the node energystability factor
331TheNode Energy Stability Factor Suppose that there are119895 intermediate nodes in the path 119875119894 119873119894 = 1198991198941 1198991198942 119899119894119895
In this paper we define the node energy stability fac-tor
ES119894119895 = 119864currrent (119894119895)119864initial (119894119895) (4)
where 119864initial(119894119895) is the initial energy of node 119895 in path 119875119894 and119864currrent(119894119895) is the current remaining energy of node 119895 in path119875119894Because a node in the path cannot be used owing to the
exhausted energy the energy stability factor of the path is theminimum energy stability factor of all nodes in the path 119875119894
ES119894 = min ES119894119895 119897119894119895 isin 119901119894 (5)
332 The Link Stability Factor In this paper we define thelink reliability factor as the remaining lifetime of the linkThe communication radius of each node is 119877 Each nodeis equipped with GPS thus every node can perceive thelocation speed time of nodes and period and send itsown coordinates and speed information to neighbor nodesAccording to the location information of nodes the remain-ing lifetime of each link can be predicted and the link stabilityfactor can be obtained As shown in Figure 2 the initialdistance between the two nodes119872 and119873 is 119889
The relative motion of the two nodes is equivalent toone node moving while the other node is stationary Thecoordinates of node119873 relative to the stationary node119872 are
Wireless Communications and Mobile Computing 5
M
N
R
MN
Figure 2 Calculate the lifetime of the link119872119873
(119909119873 minus 119909119872 119910119873 minus 119910119872) Thus the distance between119872 and119873 is119889 = radic1199091198721198732 + 1199101198721198732 According to the relative movement ofthe two nodes after time 119905 node119872 relative to position119873 is(1199091015840119872119873 1199101015840119872119873)
1199091015840119872119873 = 119909119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 cos 120579119872119873
1199101015840119872119873 = 119910119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 sin 120579119872119873 (6)
When the distance between 119872 and 119873 is 119877 the link between119872 and119873 is broken
1198891015840 = radic(1199091015840119872119873)2 + (1199101015840119872119873)2 = 119877 (7)
Take (6) into (7)
10038161003816100381610038161003816997888997888997888V119873119872100381610038161003816100381610038162 1199052 + 2 10038161003816100381610038161003816997888997888997888V119873119872
10038161003816100381610038161003816 (119909119873119872 cos 120579119873119872 + 119910119873119872 sin 120579119873119872) 119905+ 1198892 minus 1198772 = 0 (8)
where997888V119872 = (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894 + (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 997888119895 997888V119873 = (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873) 997888119894 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873) 997888119895 997888997888997888V119873119872 = 997888V119873 minus 997888V119872= (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894
+ (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 99788811989510038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic(10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)2 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872)2120579119873119872 = tanminus1(
10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 12057911987210038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)
(9)
Because 119905 cannot be negative 119905 becomes the following
119905 = radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)210038161003816100381610038161003816997888997888997888rarrV11987311987210038161003816100381610038161003816
minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
(10)
Link stability is
LS119872119873 = 119905119905max= 119905
119877 10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)2119877minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)119877
(11)
There is a 119896 link in a path between the source node 119878 and thedestination node 119863 The link stability of the path dependson the minimum link stability factor in the 119896-segment linkThus we define the link reliability factor of path 119875119894 as theminimum link stability factor
LS119894 = min LS119872119873 119897119872119873 isin 119901119894 (12)
333 The Path Reliability Factor Considering the link stabil-ity factor and node energy stability factor the comprehensivereliability factor of the path is defined as
PS119894 = 11ES119894 + 1LS119894 =ES119894 times LS119894ES119894 + LS119894
(13)
Whether the communication path between the source nodeand the destination node is stable depends on the worst linksand nodes in the path because the entire communication isinterrupted when there is a broken link or when a node exitsthe network because of exhausted battery energy After usingthe abovemethod if either of the two stability factors is smallthe value of PS119894 will be small Thus path reliability will bepoor When the source node finds multiple paths satisfyingthe QoS requirements the one with the largest PS is selectedThus PS = max PS119894 119875119894 isin 1198754 Route Discovery Process Based on JointOptimization of Dual-Channel Networks
To improve channel utilization based on the double-channelmodel and ant colony optimization algorithm the two-interlayer joint optimization routing mode is proposed Itcan increase bandwidth and make full use of idle resourcesbetween different layers In Mode 1 the routing service in thecontrol layer can only be transmitted in the control layer Ifthe control layer does not have enough channel resourcesthe service will be rejected To reduce the blocking rate of thecontrol layer joint optimization Mode 2 is proposed Whenthe data layer has enough idle resources the control packetsin the control layer can be transmitted to the data layer inreal time to realize the joint optimization of the two-layernetwork The specific routing process of the two modes is asfollows
Mode 1 It is the route discovery mode on the control layer ofDSAR
(1) Broadcast a Hello Message At the control layer eachnetwork node periodically sends a 1-hop hello message The
6 Wireless Communications and Mobile Computing
Table 1 The format of the hello message
Type Src addr Scr 119883 Scr 119884 Scr 119881 119873 energy
Table 2 The structure of neighbor
119873 addr 119873 energy LS Hops119873 119883 Nr 119884 119873 119881 Phenomenon
format of the hello message is shown in Table 1 whereldquoTyperdquo denotes packet type ldquoSrc addrrdquo is the address of thesource node that sends the hello message ldquoScr Xrdquo are the Xcoordinates of node ldquoScr Yrdquo are the Y coordinates of nodeldquoScr Vrdquo is velocity vector of node and ldquoN energyrdquo is residualenergy of node
(2) Establish the Neighbor List Each node establishes itsneighbor list by receiving hello messages sent by neighbornodes in real time In this paper each network node hasa GPS positioning device to obtain its geographic locationinformation Each network node sends hello packets period-ically and accepts those sent by neighbor nodes Each hellopacket contains node coordinates and speed information andthe neighbor list of each node The neighbor list contains thelocation vector information of all neighboring nodes Thuseach network node can obtain the location information ofneighbor nodesThe frequency of packet transmission can beset according to differentmotion scenarios and the transmis-sion frequency of hello messages can be higher for scenarioswhere network topology changes rapidly In this paper wedesign hellomessaging with a sending interval of 1 s If a nodedoes not receive neighbor hello packets in 1 s the link of thisnode is broken and the node deletes it from its neighbortable The node updates its neighbor list immediately afterthe hello message is received Otherwise the comprehensivereliability of the node is calculated according to formula (13)The format of neighbor message is shown in Table 2 whereldquo119873 addrrdquo is the address of the source node that sends thehello message ldquo119873 energyrdquo is residual energy of neighbornode ldquo119873 119883rdquo are the119883 coordinates of neighbor node ldquo119873 119884rdquoare the 119884 coordinates of neighbor node 119873 119881rdquo is velocityvector of neighbor node ldquoLSrdquo is the stability of neighbornode ldquoHopsrdquo denotes total hops of a route passing throughthis neighbor node and ldquoPhenomenonrdquo denotes the value ofpheromone
(3) Send a Forward Ant and Update Pheromone When thesource node 119878 has data packets sent to the destinationnode119863 the source node 119878 looks at the routing informationtable namely the pheromone table If there is no routinginformation and the control layer has enough bandwidththe forward ant is broadcast in the control layer If there isa route to node119863 the packet is sent directly to the data layerIn the initial stage of routing establishment the source nodebroadcasts a certain number of forward ants at the controllayer The unicast or broadcast of each intermediate nodebetween source node 119878 and destination node 119863 depends onwhether the intermediate node has pheromone If there is
Table 3 The format of forward ant message
Type Fant addr Fant Seqno TTLScr 119884 Scr 119881 119864sum Fd119896
Seqno 0Seqno 1Seqno 119899
pheromone at each intermediate node the probability thatant 119870 selects the next hop neighbor node 119895 is calculatedaccording to the following equation
119875119896119894119895 (119905)
=
[120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573
119895 isin allowed119896
0 else(14)
where 119875119896119894119895(119905) is the probability that ant 119870 selects the next hopneighbor node 119895 when node 119894moves toward the destinationnode 120591119894119895(119905) is the size of the pheromone of node 119894 at time119905 in link 119894119895 120578119894119895(119905) is a visual function of node 119894 at time 119905from node 119894 to node 119895 allowed119896 is a collection of neighbornodes of node 119894 120572 and 120573 are adjustment coefficients 120572 is arelative importance coefficient of the residual pheromone 120573is a relative importance coefficient of heuristic informationThe pheromone of node is updated as equation
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) (15)
The format of the forward ant message is shown inTable 3 where ldquoTyperdquo denotes the packet type ldquoFant addrrdquois the address of source node ldquoFant Seqnordquo is the sequencenumber of forward ants generated by source nodes ldquoSeqnordquodenotes node address visited by the forward ant ldquoTTLrdquo is thesurvival time of forward ant and ldquoFd119896rdquo denotes the numberof hops that the forward ant moves to the current node
(4) Send a Backward Ant and Update Pheromone When aforward ant arrives at the destination node it turns into abackward ant The backward ant returns to the source nodealong its former path If a fault occurs in the link of thenext hop caused by the movement of a node in the path tothe source node then the backward ant will be discardedWhen the backward ant returns to node 119894 from node 119895 thepheromone of node 119894 is updated according to the pheromoneupdate of (16)
120591119894119895 (119905 + 1) = 120591119894119895 (119905)1 minus 120588 + Δ120591119894119895 (119905) (16)
Δ120591119894119895 (119905) = 119898sum119896=1
Δ120591119896119894119895 (119905) (17)
Wireless Communications and Mobile Computing 7
Table 4 The format of backward ant message
Type Visitednode 119864min 119864avg Bd119896
Δ120591119896119894119895 (119905) = PS119894 (18)
120578119896119894119895 (119905) = 1Fd119896
(19)
The format of the backward ant message is shown inTable 4 where ldquoTyperdquo denotes the packet type ldquoVisitednoderdquodenotes the ID of node visited ldquo119864minrdquo is the energy value ofthe minimum energy node on the path that the backward antpasses through ldquo119864avgrdquo is the residual average energy of theant 119870 to the current node and ldquoBd119896rdquo denotes the number ofhops experienced by the backward ant 119870 to node 119895Mode 2 It is the route discovery process based on jointoptimization of dual-channel networks
When the control layer is congested joint optimizationMode 2 is proposed That is when the data layer has enoughidle resources the control packets in the control layer can betransmitted to the data layer in real time to realize the jointoptimization of the two-layer network The specific routingprocess of Mode 2 is as follows
(1) If the control layer has enough network resourcesthe forward ant is routed via Mode 1 in the control layerOtherwise turn to (2)
(2) When the forward ant119870 moves to node 119894 in the datatransport layer the forward ant 119870 looks at whether thereare available channel resources for ants to find paths withneighboring nodes of 119894 If not ant 119870 stops and refuses toperform the routing lookup service Otherwise it turns to (3)
(3) Ant 119870 performs the routing service in the data trans-mission layer and finds the next hop node 119895 in Mode 1
(4)When reachingnode 119895 ant119870first investigateswhetherthere are enough channel resources between node 119895 and itsneighbor nodes in the control layer to perform the routingservice If not ant 119870 continues to perform the path-findingservice in the data transport layer Otherwise ant 119870 returnsto the control layer and searches the optimal path of serviceaccording to Mode 1
5 Simulation and Analysis
In this paper to verify the reliability of the DSAR protocolNS-2 is selected and the DSAR algorithm is compared tothe EEABR algorithm [34] and the AODV algorithm AODVis a classic routing algorithm and EEABR is a successfulapplication of the ant colony algorithm in wireless ad hocnetworks
51 Simulation Setup In a wireless simulation environmenteach mobile network node is randomly distributed in the1000m times 1000m area 50 nodes are randomly arrangedaccording to the random way-point model The communi-cation radius of each node is 250m MAC layer adopts dual-channel mode The packet length is 512 b and the send rate
varies from 1 to 16 packetss The evaporation of pheromoneoccurs every 1 s The evaporation rate 120588 is set to 02 Eachvalue of 120572 and 120573 is set to 20 and 15 respectively Thesimulation time is 1000 s To reduce random errors theexperimental results will be the average of the 10 experimentsSimulation algorithm routing layers are (1) DSAR (2) AODVand (3) EEABR
52 Simulation Analysis
521 Performance Metrics for Evaluating Routing ProtocolThe performance of routing protocol is evaluated by meansof end-to-end delay average throughput packet delivery raterouting overhead and so forth The statistical methods areintroduced as follows
(1) End-to-End Delay The average end-to-end delay is thetime required from the start of routing to the end of datatransmission We calculate it with the following formula
120591 = 1119873119873sum119894=1
(119905119886119894 minus 119905119887119894) (20)
where 120591 is the average end-to-end delay 119873 is the number ofsuccessful packet transmissions 119905119886119894 is the time that packet 119894arrives at the destination node and 119905119887119894 is the time packet 119894was generated
(2) Throughput Throughput is the maximum number ofpackets that a network successfully transmits per unit time
119879 = 1119879RE minus 119879RS
119873sum119894=1
119877119887 (119894) times 8 (21)
where119879 represents the throughput119877119887(119894) represents the num-ber of bytes of packet 119894 received successfully 119873 representsthe total number of packets received from the destination119879RE represents the reception time of the data packet and 119879RSrepresents the beginning of the data packet reception
(3) Packet Delivery Rate Packet delivery rate is the ratio ofthe total number of sending packets to the total number ofreceiving packets
(4) Routing Overhead
119873 = 119875119862119875119863 (22)
where119873 represents routing overhead 119875119862 represents the totalnumber of node send control packets and 119875119863 represents thetotal number of destination node receive data packets
522 The Network Performance Varies with the Packet SendRate of the Source Node The performance of the three algo-rithms varies with the average packet sending rate of thesource node in the network as shown in Figures 3ndash6
Figure 3 shows the relationship between the average end-to-end delay and the packet delivery rate of the source nodes
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Wireless Communications and Mobile Computing 5
M
N
R
MN
Figure 2 Calculate the lifetime of the link119872119873
(119909119873 minus 119909119872 119910119873 minus 119910119872) Thus the distance between119872 and119873 is119889 = radic1199091198721198732 + 1199101198721198732 According to the relative movement ofthe two nodes after time 119905 node119872 relative to position119873 is(1199091015840119872119873 1199101015840119872119873)
1199091015840119872119873 = 119909119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 cos 120579119872119873
1199101015840119872119873 = 119910119872119873 + 10038161003816100381610038161003816997888997888997888V11987211987310038161003816100381610038161003816 119905 sin 120579119872119873 (6)
When the distance between 119872 and 119873 is 119877 the link between119872 and119873 is broken
1198891015840 = radic(1199091015840119872119873)2 + (1199101015840119872119873)2 = 119877 (7)
Take (6) into (7)
10038161003816100381610038161003816997888997888997888V119873119872100381610038161003816100381610038162 1199052 + 2 10038161003816100381610038161003816997888997888997888V119873119872
10038161003816100381610038161003816 (119909119873119872 cos 120579119873119872 + 119910119873119872 sin 120579119873119872) 119905+ 1198892 minus 1198772 = 0 (8)
where997888V119872 = (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894 + (10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 997888119895 997888V119873 = (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873) 997888119894 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873) 997888119895 997888997888997888V119873119872 = 997888V119873 minus 997888V119872= (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872) 997888119894
+ (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872) 99788811989510038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic(10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)2 + (10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 120579119872)2120579119873119872 = tanminus1(
10038161003816100381610038161003816997888V11987310038161003816100381610038161003816 sin 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 sin 12057911987210038161003816100381610038161003816997888V11987310038161003816100381610038161003816 cos 120579119873 minus 10038161003816100381610038161003816997888V11987210038161003816100381610038161003816 cos 120579119872)
(9)
Because 119905 cannot be negative 119905 becomes the following
119905 = radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)210038161003816100381610038161003816997888997888997888rarrV11987311987210038161003816100381610038161003816
minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
(10)
Link stability is
LS119872119873 = 119905119905max= 119905
119877 10038161003816100381610038161003816997888997888997888V11987311987210038161003816100381610038161003816
= radic1198772 minus (119909119873119872 sin 120579119873119872 minus 119910119873119872 cos 120579119873119872)2119877minus (119909119873119872 cos 120579119873119872 + 120579119873119872 sin 120579119873119872)119877
(11)
There is a 119896 link in a path between the source node 119878 and thedestination node 119863 The link stability of the path dependson the minimum link stability factor in the 119896-segment linkThus we define the link reliability factor of path 119875119894 as theminimum link stability factor
LS119894 = min LS119872119873 119897119872119873 isin 119901119894 (12)
333 The Path Reliability Factor Considering the link stabil-ity factor and node energy stability factor the comprehensivereliability factor of the path is defined as
PS119894 = 11ES119894 + 1LS119894 =ES119894 times LS119894ES119894 + LS119894
(13)
Whether the communication path between the source nodeand the destination node is stable depends on the worst linksand nodes in the path because the entire communication isinterrupted when there is a broken link or when a node exitsthe network because of exhausted battery energy After usingthe abovemethod if either of the two stability factors is smallthe value of PS119894 will be small Thus path reliability will bepoor When the source node finds multiple paths satisfyingthe QoS requirements the one with the largest PS is selectedThus PS = max PS119894 119875119894 isin 1198754 Route Discovery Process Based on JointOptimization of Dual-Channel Networks
To improve channel utilization based on the double-channelmodel and ant colony optimization algorithm the two-interlayer joint optimization routing mode is proposed Itcan increase bandwidth and make full use of idle resourcesbetween different layers In Mode 1 the routing service in thecontrol layer can only be transmitted in the control layer Ifthe control layer does not have enough channel resourcesthe service will be rejected To reduce the blocking rate of thecontrol layer joint optimization Mode 2 is proposed Whenthe data layer has enough idle resources the control packetsin the control layer can be transmitted to the data layer inreal time to realize the joint optimization of the two-layernetwork The specific routing process of the two modes is asfollows
Mode 1 It is the route discovery mode on the control layer ofDSAR
(1) Broadcast a Hello Message At the control layer eachnetwork node periodically sends a 1-hop hello message The
6 Wireless Communications and Mobile Computing
Table 1 The format of the hello message
Type Src addr Scr 119883 Scr 119884 Scr 119881 119873 energy
Table 2 The structure of neighbor
119873 addr 119873 energy LS Hops119873 119883 Nr 119884 119873 119881 Phenomenon
format of the hello message is shown in Table 1 whereldquoTyperdquo denotes packet type ldquoSrc addrrdquo is the address of thesource node that sends the hello message ldquoScr Xrdquo are the Xcoordinates of node ldquoScr Yrdquo are the Y coordinates of nodeldquoScr Vrdquo is velocity vector of node and ldquoN energyrdquo is residualenergy of node
(2) Establish the Neighbor List Each node establishes itsneighbor list by receiving hello messages sent by neighbornodes in real time In this paper each network node hasa GPS positioning device to obtain its geographic locationinformation Each network node sends hello packets period-ically and accepts those sent by neighbor nodes Each hellopacket contains node coordinates and speed information andthe neighbor list of each node The neighbor list contains thelocation vector information of all neighboring nodes Thuseach network node can obtain the location information ofneighbor nodesThe frequency of packet transmission can beset according to differentmotion scenarios and the transmis-sion frequency of hello messages can be higher for scenarioswhere network topology changes rapidly In this paper wedesign hellomessaging with a sending interval of 1 s If a nodedoes not receive neighbor hello packets in 1 s the link of thisnode is broken and the node deletes it from its neighbortable The node updates its neighbor list immediately afterthe hello message is received Otherwise the comprehensivereliability of the node is calculated according to formula (13)The format of neighbor message is shown in Table 2 whereldquo119873 addrrdquo is the address of the source node that sends thehello message ldquo119873 energyrdquo is residual energy of neighbornode ldquo119873 119883rdquo are the119883 coordinates of neighbor node ldquo119873 119884rdquoare the 119884 coordinates of neighbor node 119873 119881rdquo is velocityvector of neighbor node ldquoLSrdquo is the stability of neighbornode ldquoHopsrdquo denotes total hops of a route passing throughthis neighbor node and ldquoPhenomenonrdquo denotes the value ofpheromone
(3) Send a Forward Ant and Update Pheromone When thesource node 119878 has data packets sent to the destinationnode119863 the source node 119878 looks at the routing informationtable namely the pheromone table If there is no routinginformation and the control layer has enough bandwidththe forward ant is broadcast in the control layer If there isa route to node119863 the packet is sent directly to the data layerIn the initial stage of routing establishment the source nodebroadcasts a certain number of forward ants at the controllayer The unicast or broadcast of each intermediate nodebetween source node 119878 and destination node 119863 depends onwhether the intermediate node has pheromone If there is
Table 3 The format of forward ant message
Type Fant addr Fant Seqno TTLScr 119884 Scr 119881 119864sum Fd119896
Seqno 0Seqno 1Seqno 119899
pheromone at each intermediate node the probability thatant 119870 selects the next hop neighbor node 119895 is calculatedaccording to the following equation
119875119896119894119895 (119905)
=
[120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573
119895 isin allowed119896
0 else(14)
where 119875119896119894119895(119905) is the probability that ant 119870 selects the next hopneighbor node 119895 when node 119894moves toward the destinationnode 120591119894119895(119905) is the size of the pheromone of node 119894 at time119905 in link 119894119895 120578119894119895(119905) is a visual function of node 119894 at time 119905from node 119894 to node 119895 allowed119896 is a collection of neighbornodes of node 119894 120572 and 120573 are adjustment coefficients 120572 is arelative importance coefficient of the residual pheromone 120573is a relative importance coefficient of heuristic informationThe pheromone of node is updated as equation
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) (15)
The format of the forward ant message is shown inTable 3 where ldquoTyperdquo denotes the packet type ldquoFant addrrdquois the address of source node ldquoFant Seqnordquo is the sequencenumber of forward ants generated by source nodes ldquoSeqnordquodenotes node address visited by the forward ant ldquoTTLrdquo is thesurvival time of forward ant and ldquoFd119896rdquo denotes the numberof hops that the forward ant moves to the current node
(4) Send a Backward Ant and Update Pheromone When aforward ant arrives at the destination node it turns into abackward ant The backward ant returns to the source nodealong its former path If a fault occurs in the link of thenext hop caused by the movement of a node in the path tothe source node then the backward ant will be discardedWhen the backward ant returns to node 119894 from node 119895 thepheromone of node 119894 is updated according to the pheromoneupdate of (16)
120591119894119895 (119905 + 1) = 120591119894119895 (119905)1 minus 120588 + Δ120591119894119895 (119905) (16)
Δ120591119894119895 (119905) = 119898sum119896=1
Δ120591119896119894119895 (119905) (17)
Wireless Communications and Mobile Computing 7
Table 4 The format of backward ant message
Type Visitednode 119864min 119864avg Bd119896
Δ120591119896119894119895 (119905) = PS119894 (18)
120578119896119894119895 (119905) = 1Fd119896
(19)
The format of the backward ant message is shown inTable 4 where ldquoTyperdquo denotes the packet type ldquoVisitednoderdquodenotes the ID of node visited ldquo119864minrdquo is the energy value ofthe minimum energy node on the path that the backward antpasses through ldquo119864avgrdquo is the residual average energy of theant 119870 to the current node and ldquoBd119896rdquo denotes the number ofhops experienced by the backward ant 119870 to node 119895Mode 2 It is the route discovery process based on jointoptimization of dual-channel networks
When the control layer is congested joint optimizationMode 2 is proposed That is when the data layer has enoughidle resources the control packets in the control layer can betransmitted to the data layer in real time to realize the jointoptimization of the two-layer network The specific routingprocess of Mode 2 is as follows
(1) If the control layer has enough network resourcesthe forward ant is routed via Mode 1 in the control layerOtherwise turn to (2)
(2) When the forward ant119870 moves to node 119894 in the datatransport layer the forward ant 119870 looks at whether thereare available channel resources for ants to find paths withneighboring nodes of 119894 If not ant 119870 stops and refuses toperform the routing lookup service Otherwise it turns to (3)
(3) Ant 119870 performs the routing service in the data trans-mission layer and finds the next hop node 119895 in Mode 1
(4)When reachingnode 119895 ant119870first investigateswhetherthere are enough channel resources between node 119895 and itsneighbor nodes in the control layer to perform the routingservice If not ant 119870 continues to perform the path-findingservice in the data transport layer Otherwise ant 119870 returnsto the control layer and searches the optimal path of serviceaccording to Mode 1
5 Simulation and Analysis
In this paper to verify the reliability of the DSAR protocolNS-2 is selected and the DSAR algorithm is compared tothe EEABR algorithm [34] and the AODV algorithm AODVis a classic routing algorithm and EEABR is a successfulapplication of the ant colony algorithm in wireless ad hocnetworks
51 Simulation Setup In a wireless simulation environmenteach mobile network node is randomly distributed in the1000m times 1000m area 50 nodes are randomly arrangedaccording to the random way-point model The communi-cation radius of each node is 250m MAC layer adopts dual-channel mode The packet length is 512 b and the send rate
varies from 1 to 16 packetss The evaporation of pheromoneoccurs every 1 s The evaporation rate 120588 is set to 02 Eachvalue of 120572 and 120573 is set to 20 and 15 respectively Thesimulation time is 1000 s To reduce random errors theexperimental results will be the average of the 10 experimentsSimulation algorithm routing layers are (1) DSAR (2) AODVand (3) EEABR
52 Simulation Analysis
521 Performance Metrics for Evaluating Routing ProtocolThe performance of routing protocol is evaluated by meansof end-to-end delay average throughput packet delivery raterouting overhead and so forth The statistical methods areintroduced as follows
(1) End-to-End Delay The average end-to-end delay is thetime required from the start of routing to the end of datatransmission We calculate it with the following formula
120591 = 1119873119873sum119894=1
(119905119886119894 minus 119905119887119894) (20)
where 120591 is the average end-to-end delay 119873 is the number ofsuccessful packet transmissions 119905119886119894 is the time that packet 119894arrives at the destination node and 119905119887119894 is the time packet 119894was generated
(2) Throughput Throughput is the maximum number ofpackets that a network successfully transmits per unit time
119879 = 1119879RE minus 119879RS
119873sum119894=1
119877119887 (119894) times 8 (21)
where119879 represents the throughput119877119887(119894) represents the num-ber of bytes of packet 119894 received successfully 119873 representsthe total number of packets received from the destination119879RE represents the reception time of the data packet and 119879RSrepresents the beginning of the data packet reception
(3) Packet Delivery Rate Packet delivery rate is the ratio ofthe total number of sending packets to the total number ofreceiving packets
(4) Routing Overhead
119873 = 119875119862119875119863 (22)
where119873 represents routing overhead 119875119862 represents the totalnumber of node send control packets and 119875119863 represents thetotal number of destination node receive data packets
522 The Network Performance Varies with the Packet SendRate of the Source Node The performance of the three algo-rithms varies with the average packet sending rate of thesource node in the network as shown in Figures 3ndash6
Figure 3 shows the relationship between the average end-to-end delay and the packet delivery rate of the source nodes
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
6 Wireless Communications and Mobile Computing
Table 1 The format of the hello message
Type Src addr Scr 119883 Scr 119884 Scr 119881 119873 energy
Table 2 The structure of neighbor
119873 addr 119873 energy LS Hops119873 119883 Nr 119884 119873 119881 Phenomenon
format of the hello message is shown in Table 1 whereldquoTyperdquo denotes packet type ldquoSrc addrrdquo is the address of thesource node that sends the hello message ldquoScr Xrdquo are the Xcoordinates of node ldquoScr Yrdquo are the Y coordinates of nodeldquoScr Vrdquo is velocity vector of node and ldquoN energyrdquo is residualenergy of node
(2) Establish the Neighbor List Each node establishes itsneighbor list by receiving hello messages sent by neighbornodes in real time In this paper each network node hasa GPS positioning device to obtain its geographic locationinformation Each network node sends hello packets period-ically and accepts those sent by neighbor nodes Each hellopacket contains node coordinates and speed information andthe neighbor list of each node The neighbor list contains thelocation vector information of all neighboring nodes Thuseach network node can obtain the location information ofneighbor nodesThe frequency of packet transmission can beset according to differentmotion scenarios and the transmis-sion frequency of hello messages can be higher for scenarioswhere network topology changes rapidly In this paper wedesign hellomessaging with a sending interval of 1 s If a nodedoes not receive neighbor hello packets in 1 s the link of thisnode is broken and the node deletes it from its neighbortable The node updates its neighbor list immediately afterthe hello message is received Otherwise the comprehensivereliability of the node is calculated according to formula (13)The format of neighbor message is shown in Table 2 whereldquo119873 addrrdquo is the address of the source node that sends thehello message ldquo119873 energyrdquo is residual energy of neighbornode ldquo119873 119883rdquo are the119883 coordinates of neighbor node ldquo119873 119884rdquoare the 119884 coordinates of neighbor node 119873 119881rdquo is velocityvector of neighbor node ldquoLSrdquo is the stability of neighbornode ldquoHopsrdquo denotes total hops of a route passing throughthis neighbor node and ldquoPhenomenonrdquo denotes the value ofpheromone
(3) Send a Forward Ant and Update Pheromone When thesource node 119878 has data packets sent to the destinationnode119863 the source node 119878 looks at the routing informationtable namely the pheromone table If there is no routinginformation and the control layer has enough bandwidththe forward ant is broadcast in the control layer If there isa route to node119863 the packet is sent directly to the data layerIn the initial stage of routing establishment the source nodebroadcasts a certain number of forward ants at the controllayer The unicast or broadcast of each intermediate nodebetween source node 119878 and destination node 119863 depends onwhether the intermediate node has pheromone If there is
Table 3 The format of forward ant message
Type Fant addr Fant Seqno TTLScr 119884 Scr 119881 119864sum Fd119896
Seqno 0Seqno 1Seqno 119899
pheromone at each intermediate node the probability thatant 119870 selects the next hop neighbor node 119895 is calculatedaccording to the following equation
119875119896119894119895 (119905)
=
[120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573sum119895isinallowed119896 [120591119894119895 (119905)]120572 [120578119894119895 (119905)]120573
119895 isin allowed119896
0 else(14)
where 119875119896119894119895(119905) is the probability that ant 119870 selects the next hopneighbor node 119895 when node 119894moves toward the destinationnode 120591119894119895(119905) is the size of the pheromone of node 119894 at time119905 in link 119894119895 120578119894119895(119905) is a visual function of node 119894 at time 119905from node 119894 to node 119895 allowed119896 is a collection of neighbornodes of node 119894 120572 and 120573 are adjustment coefficients 120572 is arelative importance coefficient of the residual pheromone 120573is a relative importance coefficient of heuristic informationThe pheromone of node is updated as equation
120591119894119895 (119905 + 1) = (1 minus 120588) 120591119894119895 (119905) (15)
The format of the forward ant message is shown inTable 3 where ldquoTyperdquo denotes the packet type ldquoFant addrrdquois the address of source node ldquoFant Seqnordquo is the sequencenumber of forward ants generated by source nodes ldquoSeqnordquodenotes node address visited by the forward ant ldquoTTLrdquo is thesurvival time of forward ant and ldquoFd119896rdquo denotes the numberof hops that the forward ant moves to the current node
(4) Send a Backward Ant and Update Pheromone When aforward ant arrives at the destination node it turns into abackward ant The backward ant returns to the source nodealong its former path If a fault occurs in the link of thenext hop caused by the movement of a node in the path tothe source node then the backward ant will be discardedWhen the backward ant returns to node 119894 from node 119895 thepheromone of node 119894 is updated according to the pheromoneupdate of (16)
120591119894119895 (119905 + 1) = 120591119894119895 (119905)1 minus 120588 + Δ120591119894119895 (119905) (16)
Δ120591119894119895 (119905) = 119898sum119896=1
Δ120591119896119894119895 (119905) (17)
Wireless Communications and Mobile Computing 7
Table 4 The format of backward ant message
Type Visitednode 119864min 119864avg Bd119896
Δ120591119896119894119895 (119905) = PS119894 (18)
120578119896119894119895 (119905) = 1Fd119896
(19)
The format of the backward ant message is shown inTable 4 where ldquoTyperdquo denotes the packet type ldquoVisitednoderdquodenotes the ID of node visited ldquo119864minrdquo is the energy value ofthe minimum energy node on the path that the backward antpasses through ldquo119864avgrdquo is the residual average energy of theant 119870 to the current node and ldquoBd119896rdquo denotes the number ofhops experienced by the backward ant 119870 to node 119895Mode 2 It is the route discovery process based on jointoptimization of dual-channel networks
When the control layer is congested joint optimizationMode 2 is proposed That is when the data layer has enoughidle resources the control packets in the control layer can betransmitted to the data layer in real time to realize the jointoptimization of the two-layer network The specific routingprocess of Mode 2 is as follows
(1) If the control layer has enough network resourcesthe forward ant is routed via Mode 1 in the control layerOtherwise turn to (2)
(2) When the forward ant119870 moves to node 119894 in the datatransport layer the forward ant 119870 looks at whether thereare available channel resources for ants to find paths withneighboring nodes of 119894 If not ant 119870 stops and refuses toperform the routing lookup service Otherwise it turns to (3)
(3) Ant 119870 performs the routing service in the data trans-mission layer and finds the next hop node 119895 in Mode 1
(4)When reachingnode 119895 ant119870first investigateswhetherthere are enough channel resources between node 119895 and itsneighbor nodes in the control layer to perform the routingservice If not ant 119870 continues to perform the path-findingservice in the data transport layer Otherwise ant 119870 returnsto the control layer and searches the optimal path of serviceaccording to Mode 1
5 Simulation and Analysis
In this paper to verify the reliability of the DSAR protocolNS-2 is selected and the DSAR algorithm is compared tothe EEABR algorithm [34] and the AODV algorithm AODVis a classic routing algorithm and EEABR is a successfulapplication of the ant colony algorithm in wireless ad hocnetworks
51 Simulation Setup In a wireless simulation environmenteach mobile network node is randomly distributed in the1000m times 1000m area 50 nodes are randomly arrangedaccording to the random way-point model The communi-cation radius of each node is 250m MAC layer adopts dual-channel mode The packet length is 512 b and the send rate
varies from 1 to 16 packetss The evaporation of pheromoneoccurs every 1 s The evaporation rate 120588 is set to 02 Eachvalue of 120572 and 120573 is set to 20 and 15 respectively Thesimulation time is 1000 s To reduce random errors theexperimental results will be the average of the 10 experimentsSimulation algorithm routing layers are (1) DSAR (2) AODVand (3) EEABR
52 Simulation Analysis
521 Performance Metrics for Evaluating Routing ProtocolThe performance of routing protocol is evaluated by meansof end-to-end delay average throughput packet delivery raterouting overhead and so forth The statistical methods areintroduced as follows
(1) End-to-End Delay The average end-to-end delay is thetime required from the start of routing to the end of datatransmission We calculate it with the following formula
120591 = 1119873119873sum119894=1
(119905119886119894 minus 119905119887119894) (20)
where 120591 is the average end-to-end delay 119873 is the number ofsuccessful packet transmissions 119905119886119894 is the time that packet 119894arrives at the destination node and 119905119887119894 is the time packet 119894was generated
(2) Throughput Throughput is the maximum number ofpackets that a network successfully transmits per unit time
119879 = 1119879RE minus 119879RS
119873sum119894=1
119877119887 (119894) times 8 (21)
where119879 represents the throughput119877119887(119894) represents the num-ber of bytes of packet 119894 received successfully 119873 representsthe total number of packets received from the destination119879RE represents the reception time of the data packet and 119879RSrepresents the beginning of the data packet reception
(3) Packet Delivery Rate Packet delivery rate is the ratio ofthe total number of sending packets to the total number ofreceiving packets
(4) Routing Overhead
119873 = 119875119862119875119863 (22)
where119873 represents routing overhead 119875119862 represents the totalnumber of node send control packets and 119875119863 represents thetotal number of destination node receive data packets
522 The Network Performance Varies with the Packet SendRate of the Source Node The performance of the three algo-rithms varies with the average packet sending rate of thesource node in the network as shown in Figures 3ndash6
Figure 3 shows the relationship between the average end-to-end delay and the packet delivery rate of the source nodes
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Wireless Communications and Mobile Computing 7
Table 4 The format of backward ant message
Type Visitednode 119864min 119864avg Bd119896
Δ120591119896119894119895 (119905) = PS119894 (18)
120578119896119894119895 (119905) = 1Fd119896
(19)
The format of the backward ant message is shown inTable 4 where ldquoTyperdquo denotes the packet type ldquoVisitednoderdquodenotes the ID of node visited ldquo119864minrdquo is the energy value ofthe minimum energy node on the path that the backward antpasses through ldquo119864avgrdquo is the residual average energy of theant 119870 to the current node and ldquoBd119896rdquo denotes the number ofhops experienced by the backward ant 119870 to node 119895Mode 2 It is the route discovery process based on jointoptimization of dual-channel networks
When the control layer is congested joint optimizationMode 2 is proposed That is when the data layer has enoughidle resources the control packets in the control layer can betransmitted to the data layer in real time to realize the jointoptimization of the two-layer network The specific routingprocess of Mode 2 is as follows
(1) If the control layer has enough network resourcesthe forward ant is routed via Mode 1 in the control layerOtherwise turn to (2)
(2) When the forward ant119870 moves to node 119894 in the datatransport layer the forward ant 119870 looks at whether thereare available channel resources for ants to find paths withneighboring nodes of 119894 If not ant 119870 stops and refuses toperform the routing lookup service Otherwise it turns to (3)
(3) Ant 119870 performs the routing service in the data trans-mission layer and finds the next hop node 119895 in Mode 1
(4)When reachingnode 119895 ant119870first investigateswhetherthere are enough channel resources between node 119895 and itsneighbor nodes in the control layer to perform the routingservice If not ant 119870 continues to perform the path-findingservice in the data transport layer Otherwise ant 119870 returnsto the control layer and searches the optimal path of serviceaccording to Mode 1
5 Simulation and Analysis
In this paper to verify the reliability of the DSAR protocolNS-2 is selected and the DSAR algorithm is compared tothe EEABR algorithm [34] and the AODV algorithm AODVis a classic routing algorithm and EEABR is a successfulapplication of the ant colony algorithm in wireless ad hocnetworks
51 Simulation Setup In a wireless simulation environmenteach mobile network node is randomly distributed in the1000m times 1000m area 50 nodes are randomly arrangedaccording to the random way-point model The communi-cation radius of each node is 250m MAC layer adopts dual-channel mode The packet length is 512 b and the send rate
varies from 1 to 16 packetss The evaporation of pheromoneoccurs every 1 s The evaporation rate 120588 is set to 02 Eachvalue of 120572 and 120573 is set to 20 and 15 respectively Thesimulation time is 1000 s To reduce random errors theexperimental results will be the average of the 10 experimentsSimulation algorithm routing layers are (1) DSAR (2) AODVand (3) EEABR
52 Simulation Analysis
521 Performance Metrics for Evaluating Routing ProtocolThe performance of routing protocol is evaluated by meansof end-to-end delay average throughput packet delivery raterouting overhead and so forth The statistical methods areintroduced as follows
(1) End-to-End Delay The average end-to-end delay is thetime required from the start of routing to the end of datatransmission We calculate it with the following formula
120591 = 1119873119873sum119894=1
(119905119886119894 minus 119905119887119894) (20)
where 120591 is the average end-to-end delay 119873 is the number ofsuccessful packet transmissions 119905119886119894 is the time that packet 119894arrives at the destination node and 119905119887119894 is the time packet 119894was generated
(2) Throughput Throughput is the maximum number ofpackets that a network successfully transmits per unit time
119879 = 1119879RE minus 119879RS
119873sum119894=1
119877119887 (119894) times 8 (21)
where119879 represents the throughput119877119887(119894) represents the num-ber of bytes of packet 119894 received successfully 119873 representsthe total number of packets received from the destination119879RE represents the reception time of the data packet and 119879RSrepresents the beginning of the data packet reception
(3) Packet Delivery Rate Packet delivery rate is the ratio ofthe total number of sending packets to the total number ofreceiving packets
(4) Routing Overhead
119873 = 119875119862119875119863 (22)
where119873 represents routing overhead 119875119862 represents the totalnumber of node send control packets and 119875119863 represents thetotal number of destination node receive data packets
522 The Network Performance Varies with the Packet SendRate of the Source Node The performance of the three algo-rithms varies with the average packet sending rate of thesource node in the network as shown in Figures 3ndash6
Figure 3 shows the relationship between the average end-to-end delay and the packet delivery rate of the source nodes
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
8 Wireless Communications and Mobile Computing
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetssecond)
0
02
04
06
08
1
12
14
Del
ay (s
econ
d)
Figure 3 End-to-end delay
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsseconds)
times104
0
1
2
3
4
5
6
7
8
Thro
ughp
ut (b
ps)
Figure 4 Throughput
in the three algorithms As shown in the figure the averageend-to-end delay of the three algorithms increases with theincrease of the sending speed of the source nodeThe averageend-to-end delay of DSAR is significantly smaller than thatof EEABR and AODV This is because with the increasingpacket sending rate of the source node and the congestionof the network the DSAR selects the nodes having large
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
20
30
40
50
60
70
80
90
100
Pack
et D
elive
ry ra
te (
)
Figure 5 Packet delivery rate
DSARAODV + DUAL-CHANNELEEABR + DUAL-CHANNEL
2 4 6 8 10 12 14 160Packet Send Rate (packetsecond)
0
5
10
15
20
25
30
35
40
Rout
ing
load
Figure 6 Routing overhead
stability and large residual power to transmit data packetsThis reduces the delay caused by link interruption and routingrestart repair Simulation results show that compared withthe classical EEABR and AODV algorithms the average end-to-end delay of the RSAR algorithm is reduced
Figure 4 shows the relationship between the throughputof the three algorithms and the packet delivery rate of thesource node As can be seen in the graph the throughput
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Wireless Communications and Mobile Computing 9
of each routing discovery increases with the increase ofthe packet delivery rate of the source node from the threealgorithms The routing discovery throughput of DSAR issignificantly higher than EEABR and AODV This is becausethe dual-channel mechanism is adopted by DSAR to sepa-rate control packets from data packets which reduces thechannel switching and data collision probability DSAR usesa comprehensive stability prediction mechanism to select thepath having good stability and fewer hops and establisheshigh-quality routingwhich reduces the probability of routingrestarts and improves throughput
Figure 5 shows the relationship between the packet deliv-ery rate of the three algorithms and the packet delivery rate ofthe source node From the graph it can be seen that with theincrease of routing load the packet delivery rate of DSAR ishigher than that of AODV and EEABR However DSAR andEEABR decrease rapidly with the increase of packet sendingrate whereas AODV remains unchanged DSAR has packetdelivery rates higher thanAODVand EEABRThis is becausethe control packets and the data packets are transmitted overdifferent channels which reduce packet collision and increasenetwork bandwidth The poor performance of EEABR iscaused by the increase of transmission packet collisions andthe periodic transmission of ant packets
Figure 6 shows the relationship between the routing over-head of three algorithms and the packet delivery rate of sourcenodes From the simulation results the routing overheadis reduced with the increased packet sending rate EEABRgenerates a large number of ant packets which increases thecost of route discovery However the overhead of ADOV islower than that of EEABR and DSAR because AODV useson-demand routing DSAR needs to send periodic probepackets to find stable nodes and links so the cost of DSARwill be slightly higher than EEABRWith the increase of loadthe routing overhead of DSAR approaches that of AODVbecause the frequent retransmission caused by the instabilityin AODV leads to the increase of routing overhead
6 Conclusion
To improve the reliability of routing protocol in wirelessad hoc networks a reliable ant colony algorithm for dual-channel systems was proposed In the DSAR algorithm thedouble-layer mechanism of control layer and data layerseparation was established which reduced packet collisionand channel handoff delay and increased network band-width Simultaneously when the data layer had enough idleresources it transferred the blocked routing service overthe control layer to the data layer in real time completingthe joint scheduling of the double-layer network and reduc-ing the congestion rate Moreover the reliability predictionmechanism was proposed which enhanced link reliabilityand reduced the probability of routing restart Also for thedynamic change of topology in ad hoc networks the antcolony algorithm was used to adapt the dynamic changes ofnetwork topology The comprehensive reliability value of theproposed reliability prediction model was used as one of thebases of pheromone updates for the ant colony algorithmSimulation results show that compared with the classic
AODV and EEABR models DSAR improved the reliabilityof routing protocols
Data Availability
The data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This work was supported by the National Natural ScienceFoundation of China (Grant no 61601475)
References
[1] F Khan Q Jabeen S Khan and M Ahmad PerformanceImprovement inMultihopWirelessMobile AdhocNetworks 2016
[2] G Li L Boukhatem and J Wu ldquoAdaptive Quality-of-Service-BasedRouting forVehicularAdHocNetworkswithAntColonyOptimizationrdquo IEEE Transactions on Vehicular Technology vol66 no 4 pp 3249ndash3264 2017
[3] C Perkins E Belding-Royer and S Das ldquoAd hoc on-demanddistance vector (AODV) routingrdquo No RFC 3561 2003
[4] T H Clausen and A C D Verdiere ldquoThe LLN On-demandAd hoc Distance-vector Routing Protocol -Next Generation(LOADngrdquo in Networking amp Internet Architecture 2015
[5] D U Chuan-Bao H D Quan L I Zhao-Rui and P ZCui ldquoDesign and Analysis of Hierarchical Routing Protocolfor Wireless Dual-Channel Ad Hoc Networksrdquo in Journal ofCommand amp Control 2015
[6] G Pei M Gerla X Hong and C-C Chiang ldquoA wireless hier-archical routing protocol with groupmobilityrdquo in Proceedings ofthe IEEE Wireless Communications and Networking Conferencepp 1538ndash1542 IEEE New Orleans La USA September 1999
[7] J J Garcia-Luna-Aceves and M Spohn ldquoSource-tree routingin wireless networksrdquo in Proceedings of the 7th InternationalConference on Network Protocols (ICNP rsquo99) pp 273ndash282 IEEENovember 1999
[8] S Murthy and J J Garcia-Luna-Aceves ldquoRouting protocolfor packet radio networksrdquo in Proceedings of the 1st AnnualInternational Conference on Mobile Computing and Networking(MobiCom rsquo95) pp 86ndash95 Berkeley Calif USA November1995
[9] W Guo J Li G Chen Y Niu and C Chen ldquoA PSO-OptimizedReal-Time Fault-Tolerant Task Allocation Algorithm in Wire-less Sensor Networksrdquo IEEE Transactions on Parallel and Dis-tributed Systems vol 26 no 12 pp 3236ndash3249 2015
[10] X Luo D Zhang L T Yang J Liu X Chang and H Ning ldquoAkernelmachine-based secure data sensing and fusion scheme inwireless sensor networks for the cyber-physical systemsrdquo FutureGeneration Computer Systems vol 61 pp 85ndash96 2016
[11] C E Perkins and P Bhagwat ldquoHighly dynamic destination-sequenced distance-vector routing (DSDV) formobile comput-ersrdquo Computer Communication Review vol 24 no 4 pp 234ndash244 1994
[12] M Gerls Fisheye State Routing (FSR) for Ad Hoc NetworksInternet Draft draft-ietf-manet-fsr-03txt 2002
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
10 Wireless Communications and Mobile Computing
[13] B Xu and F Sun ldquoComposite intelligent learning control ofstrict-feedback systemswith disturbancerdquo IEEETransactions onCybernetics vol PP no 99 pp 1ndash12 2017
[14] Tsu-Wei Chen and M Gerla ldquoGlobal state routing a newrouting scheme for ad-hoc wireless networksrdquo in Proceedings ofthe ICC rsquo98 1998 IEEE International Conference on Communica-tions Conference Record pp 171ndash175 Atlanta GA USA 1998
[15] D B Johnson ldquoThe dynamic source routing in ad hoc wirelessnetworks (DSR)rdquo inMobile Computing 1996
[16] G Aggelou and R Tafazolli ldquoRDMAR A bandwidth-efficientrouting protocol for mobile ad hoc networksrdquo in Proceedingsof the 2nd ACM International Workshop on Wireless MobileMultimedia WOWMOM 1999 pp 26ndash33 usa
[17] B K Young and H V Nitin ldquoLocation-Aided Routing (LAR)in mobile ad hoc networksrdquoWireless Networks vol 6 no 4 pp307ndash321 2000
[18] L Barolli Y Honma A Koyama A Durresi and J Arai ldquoAselective border-casting zone routing protocol for ad-hoc net-worksrdquo in Proceedings of the 15th International Workshop onDatabase and Expert Systems Applications pp 326ndash330September 2004
[19] K Yang and J-F Ma ldquoHybrid wireless mesh protocolrdquo TongxinXuebaoJournal on Communication vol 30 no 11 A pp 133ndash139 2009
[20] S Wu X Tan and S Jia ldquoAOHR AODV and OLSR hybridrouting protocol for mobile ad hoc networksrdquo in Proceedings ofthe 2006 International Conference on Communications Circuitsand Systems ICCCAS pp 1487ndash1491 chn June 2006
[21] S Kashef and H Nezamabadi-pour ldquoAn advanced ACO algo-rithm for feature subset selectionrdquoNeurocomputing vol 147 no1 pp 271ndash279 2015
[22] A George Performance Analysis of Energy Efficient LocationBasedACORoutingAlgorithm forMobile AdHocNetworks usingBonn Motion Mobility Models 2015
[23] G Di Caro and M Dorigo ldquoAntNet Distributed stigmergeticcontrol for communications networksrdquo Journal of ArtificialIntelligence Research vol 9 pp 317ndash365 2011
[24] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 British Columbia Canada August 2002
[25] F Correia andTVazao ldquoSimple ant routing algorithm strategiesfor a (Multipurpose) MANET modelrdquo Ad Hoc Networks vol 8no 8 pp 810ndash823 2010
[26] G Di Caro F Ducatelle and L M Gambardella ldquoAntHocNetan adaptive nature-inspired algorithm for routing in mobile adhoc networksrdquo European Transactions on Telecommunicationsvol 16 no 5 pp 443ndash455 2005
[27] J Zhou H Tan Y Deng L Cui and D D Liu ldquoAnt colony-based energy control routing protocol for mobile ad hocnetworks under different node mobility modelsrdquo EURASIPJournal on Wireless Communications and Networking vol 2016no 1 article no 105 2016
[28] D Kadono T Izumi F Ooshita H Kakugawa and T Masu-zawa ldquoAn ant colony optimization routing based on robustnessfor ad hoc networks with GPSsrdquo Ad Hoc Networks vol 8 no 1pp 63ndash76 2010
[29] K H Li and J S Leu Weakly connected dominating set-assisted ant-based routing protocol for wireless ad-hoc networksPergamon Press Inc 2015
[30] S Misra S K Dhurandher M S Obaidat K Verma and PGupta ldquoA low-overhead fault-tolerant routing algorithm formobile ad hoc networks A scheme and its simulation analysisrdquoSimulationModelling Practice andTheory vol 18 no 5 pp 637ndash649 2010
[31] Y Xue and K Nahrstedt ldquoFault tolerant routing in mobile adhoc networksrdquo in Proceedings of the 2003 IEEE Wireless Com-munications and Networking ConferenceThe Dawn of PervasiveCommunication WCNC 2003 pp 1174ndash1179 usa March 2003
[32] S Kamali and J Opatrny ldquoPOSANT a position Based AntColony Routing Algorithm for Mobile Ad-hoc NetworksrdquoJournal of Networks vol 3 21 pages 2008
[33] S Rathore and M R Khan ldquoEnhance congestion control mul-tipath routing with ANT optimization in Mobile ad hocNetworkrdquo in Proceedings of the 2016 International Conference onICT in Business Industry and Government ICTBIG 2016 indNovember 2016
[34] I Woungang M S Obaidat S K Dhurandher A Ferwornand W Shah ldquoAn ant-swarm inspired energy-efficient ad hocon-demand routing protocol for mobile ad hoc networksrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 3645ndash3649 Budapest Hungary June 2013
[35] A Biradar R C Thool R Velur and T S Indumathi ldquoDualchannel based multi-objectives genetic routing protocol for ad-hoc networks and optical networks using power aware clusteredtopologyrdquo in Proceedings of the International Conference onOptical Engineering pp 1ndash6 2013
[36] K Liu S Liu and H Jiao ldquoRouting algorithm based on antcolony optimization in the dual-channel wireless sensor net-workrdquo Journal of Xidian University vol 40 pp 58ndash62 2013
[37] I Alaya C Solnon and K Ghedira ldquoAnt Colony Optimizationfor Multi-Objective Optimization Problemsrdquo in Proceedings ofthe IEEE International Conference on TOOLS with ArtificialIntelligence pp 450ndash457 2017
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom